A method, apparatus and equipment for vehicle competition analysis

Abstract

This application provides a vehicle competition analysis method, apparatus, and equipment, relating to the field of data analysis technology. The method includes: acquiring vehicle replacement and repurchase sales data within a target time period; calculating a relative preference index between the preceding and following vehicles based on the category identifiers of the preceding and following vehicles in the target dimension, generating a relative preference matrix under the target dimension; calculating the price similarity among all preceding vehicles based on their reference prices in the replacement and repurchase sales data, generating a preceding vehicle similarity matrix; and obtaining the competition intensity value between the target vehicle and other vehicles under the target dimension based on the relative preference matrix and the preceding vehicle similarity matrix. This application can improve the accuracy of vehicle competition analysis.

Description

Technical Field

[0001] This application relates to the field of data analysis technology, and more specifically, to a method, apparatus, and equipment for vehicle competition analysis. Background Technology

[0002] In the context of increasingly fierce competition in the automotive market, replacement purchases have become the core driver of market growth, and automakers' competitive focus has shifted from first-time car buyers to competing for existing customers of other brands. Accurately identifying and quantifying competitive relationships between brands or models has become a strategic requirement for formulating product strategies and marketing decisions. Currently, automakers' analysis of competitor relationships relies heavily on subjective experience, questionnaires, and manual parameter comparisons, lacking the support of real-world user replacement behavior big data, making it difficult to objectively quantify the competitive intensity between models and brands. Traditional methods yield highly subjective results with poor reproducibility, failing to accurately identify core competitors, the evolution of competitive relationships, and the competitive landscape of segmented markets, easily leading to market misjudgments and resource misallocation.

[0003] Therefore, there is an urgent need for a quantifiable and objective analysis method based on big data to overcome the limitations of traditional experience-based judgment, provide reliable data support for automakers' product planning, marketing strategy formulation and resource optimization, and meet the market analysis needs of the era of stock competition. Summary of the Invention

[0004] This application provides a vehicle competition analysis method, apparatus, and device that can improve the accuracy of vehicle competition analysis.

[0005] In a first aspect, embodiments of this application provide a vehicle competition analysis method, the method comprising: Obtain vehicle replacement and purchase sales data within a target time period; wherein, the replacement and purchase sales data includes at least: the category identifier of the preceding vehicle in at least one dimension and its reference price, and the category identifier of the following vehicle in at least one dimension and its reference price; Based on the category identifier of the preceding vehicle in the target dimension and the category identifier of the following vehicle in the target dimension in the sales data of the replacement purchase, calculate the relative preference index between the preceding vehicle and the following vehicle, and generate the relative preference matrix under the target dimension. Based on the reference price of the preceding vehicle in the sales data of the replacement purchase, calculate the price similarity between all the preceding vehicle objects and generate a preceding vehicle similarity matrix; Based on the relative preference matrix and the preceding vehicle similarity matrix, the competition intensity value between the target vehicle object and other vehicle objects in the target dimension is obtained.

[0006] Optionally, the step of calculating the relative preference index between the preceding vehicle and the following vehicle in the target dimension based on the category identifier of the preceding vehicle in the target dimension and the category identifier of the following vehicle in the target dimension, and generating a relative preference matrix in the target dimension, includes: Based on the sales data of the replacement purchase, vehicle pairs are grouped to obtain multiple replacement purchase vehicle pairs; Based on the category identifier of the preceding vehicle in the target dimension and the category identifier of the preceding vehicle in the target dimension in each vehicle replacement / replacement pair, calculate the relative preference index between the preceding vehicle and the following vehicle in the vehicle replacement / replacement pair to obtain the relative preference index of the vehicle replacement / replacement pair. A relative preference matrix under the target dimension is generated based on the relative preference indices of multiple vehicle replacement / replacement purchase pairs.

[0007] Optionally, the step of calculating the relative preference index between the preceding and following vehicles in each vehicle replacement / replacement pair based on the category identifier of the preceding vehicle in the target dimension and the category identifier of the preceding vehicle in the target dimension, to obtain the relative preference index of the vehicle replacement / replacement pair, includes: Based on the category identifier of the preceding vehicle in the target dimension among multiple vehicle replacement and upgrade pairs, and the category identifier of the preceding vehicle in the target dimension, obtain the total number of people who have upgraded the preceding vehicle and the total number of vehicles that have upgraded to the following vehicle. Based on the total number of people who traded in the preceding vehicle and the total number of vehicles traded in from the preceding vehicle to the following vehicle in the vehicle pair, calculate the conditional probability of trading in to the following vehicle. The marginal probability of acquiring a vehicle is calculated based on the total number of vehicles acquired through trade-in and the total number of vehicles acquired through trade-in in the vehicle pair. Based on the conditional probability and marginal probability of acquiring the next vehicle, the relative preference index between the previous vehicle and the next vehicle in the vehicle replacement pair is calculated to obtain the relative preference index of the vehicle replacement pair.

[0008] Optionally, the method further includes: Based on the number of the vehicle replacement / replacement pairs and the preset relative preference index, the relative preference index of each vehicle replacement / replacement pair is optimized.

[0009] Optionally, the step of calculating the price similarity among all the preceding vehicles based on the reference price of the preceding vehicle in the replacement purchase sales data, and generating a preceding vehicle similarity matrix, includes: Based on the reference price of the preceding vehicle in the sales data of the replacement purchase, obtain the number of vehicles of the preceding vehicle in each preset price range; Based on the total number of preceding vehicles and the number of vehicles in each preset price range, obtain the probability distribution of the preceding vehicles relative to each preset price range. Calculate the price distribution distance between different preceding vehicles based on the probability distribution of the different preceding vehicles; Calculate the price similarity between different preceding objects based on the price distribution distance between different preceding objects; The preceding vehicle similarity matrix is ​​generated based on the price similarity between multiple sets of different preceding vehicle objects.

[0010] Optionally, calculating the price similarity between different preceding vehicles based on the price distribution distance between different preceding vehicles includes: By using a preset attenuation coefficient, the price distribution distance between different preceding vehicle objects is mapped to obtain the price similarity between different preceding vehicle objects.

[0011] Optionally, obtaining the competition intensity value between the target vehicle object and other vehicle objects in the target dimension based on the relative preference matrix and the preceding vehicle similarity matrix includes: Based on the relative preference matrix, obtain the first relative preference coefficient of the target vehicle object relative to the corresponding preceding vehicle object, and the second relative preference coefficients of other vehicle objects relative to the corresponding preceding vehicle object; Based on the first relative preference coefficient of the target vehicle object and the market share of the corresponding preceding vehicle object, obtain the first vector of the target vehicle object; Based on the second relative preference coefficient of the other vehicle objects and the market share of the corresponding preceding vehicle object, obtain the second vector of the other vehicle objects; Based on the first vector of the target vehicle object, the second vector of the other vehicle objects, and the similarity matrix of the preceding vehicle, the competition intensity value between the target vehicle object and the other vehicle objects in the target dimension is obtained.

[0012] Optionally, obtaining the competition intensity value between the target vehicle object and other vehicle objects in the target dimension based on the first vector of the target vehicle object, the second vector of the other vehicle objects, and the preceding vehicle similarity matrix includes: The dot product vector of the target vehicle object and the other vehicle objects is obtained by performing a dot product on the first vector of the target vehicle object, the second vector of the other vehicle objects, and the similarity matrix of the preceding vehicle. Based on the preceding vehicle similarity matrix, the vector norms of the first vector of the target vehicle object and the second vector of the other vehicle objects are calculated respectively to obtain the first vector norm and the second vector norm; Based on the dot product vector of the target vehicle object and the other vehicle objects, as well as the first vector norm and the second vector norm, the competition intensity value between the target vehicle object and the other vehicle objects in the target dimension is obtained.

[0013] Secondly, embodiments of this application also provide a vehicle competition analysis device, the device comprising: The acquisition module is used to acquire vehicle replacement and purchase sales data within a target time period; wherein, the replacement and purchase sales data includes at least: the category identifier of the preceding vehicle in at least one dimension and its reference price, and the category identifier of the following vehicle in at least one dimension and its reference price. The calculation module is used to calculate the relative preference index between the preceding vehicle and the following vehicle in the target dimension based on the category identifier of the preceding vehicle in the replacement purchase sales data and the category identifier of the following vehicle in the target dimension, and generate a relative preference matrix in the target dimension; calculate the price similarity between all preceding vehicle objects based on the reference price of the preceding vehicle in the replacement purchase sales data, and generate a preceding vehicle similarity matrix; and obtain the competition intensity value between the target vehicle object and other vehicle objects in the target dimension based on the relative preference matrix and the preceding vehicle similarity matrix.

[0014] Thirdly, embodiments of this application also provide an electronic device, including: a processor, a memory, and a bus, wherein the memory stores program instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and the processor executes the program instructions to perform the steps of the vehicle competition analysis method as described in any of the first aspects.

[0015] This application provides a vehicle competition analysis method, apparatus, and equipment. First, it acquires vehicle replacement and purchase sales data within a target time period. The data must include at least one dimension for distinguishing vehicle categories, identifying preceding and following vehicles, and their respective reference prices. Based on this data, it first calculates the relative preference index for users replacing their vehicles with those of the preceding vehicle under a selected target dimension, and then summarizes this to form a relative preference matrix for that dimension. Next, it uses the reference prices of all preceding vehicles in the data to calculate the price distribution similarity between them, thereby constructing a preceding vehicle similarity matrix. Finally, it combines the aforementioned calculated relative preference matrix and preceding vehicle similarity matrix to obtain the competition intensity value between the target vehicle and other vehicle objects under the target dimension. By applying this method, massive amounts of real replacement and upgrade purchase data can be transformed into objective and quantifiable competitive intensity values. This allows automakers to clearly understand which products directly compete with their own products in specific dimensions, as well as the relative strength of the competition. Based on the competitive intensity values, companies can improve the scientific nature and accuracy of their decisions when formulating market strategies, allocating marketing resources, and evaluating product positioning, effectively responding to market competition and significantly enhancing the scientific nature, accuracy, and timeliness of market analysis and strategic planning. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating a vehicle competition analysis method provided in an embodiment of this application; Figure 2 A flowchart illustrating the generation of a relative preference matrix in a vehicle competition analysis method provided in this application embodiment; Figure 3 A flowchart illustrating the generation of a relative preference matrix in another vehicle competition analysis method provided in this application embodiment; Figure 4 A schematic diagram illustrating the process of generating a preceding vehicle similarity matrix in a vehicle competition analysis method provided in this application embodiment; Figure 5 This is a schematic diagram of the process for obtaining the competition intensity value in a vehicle competition analysis method provided in an embodiment of this application; Figure 6 A flowchart illustrating the process of obtaining the competition intensity value in another vehicle competition analysis method provided in this application embodiment; Figure 7A schematic diagram of a vehicle competition analysis device provided in an embodiment of this application; Figure 8 This is a schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0019] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0020] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0021] Before providing a detailed explanation of this application, let's first introduce its application scenarios.

[0022] In the competitive landscape of the existing automotive market, automakers need to identify competitors across different segments and develop targeted product and marketing strategies. For example, in the new energy vehicle segment, automakers need to understand the competitive intensity between their own models and competing products in terms of user replacement preferences. However, existing competitive analysis methods largely rely on sales statistics or subjective parameter comparisons, failing to reflect the competitive relationships behind actual user replacement behavior. This makes accurate analysis difficult, resulting in a lack of scientific basis for automakers' competitive strategy formulation and inefficient resource allocation.

[0023] Based on this, this application provides a vehicle competition analysis method, apparatus, and equipment. First, it acquires vehicle replacement and upgrade sales data within a target time period. The data must include at least one dimension to distinguish preceding and following vehicle categories, along with their respective reference prices. Based on this data, it first calculates the relative preference index for users upgrading from preceding to following vehicles under a selected target dimension, and summarizes this to form a relative preference matrix for that dimension. Then, it uses the reference prices of all preceding vehicles in the data to calculate the price distribution similarity between them, thereby constructing a preceding vehicle similarity matrix. Finally, it comprehensively processes the aforementioned relative preference matrix and preceding vehicle similarity matrix to obtain the competitive intensity value between the target vehicle and other vehicle categories under the target dimension. This allows automakers to clearly understand which products directly compete with their own products under a specific dimension, and the relative strength of that competition. Based on the competitive intensity value, companies can improve the scientific and accurate nature of their decisions when formulating market strategies, allocating marketing resources, and evaluating product positioning, effectively responding to market competition and significantly improving the scientific, accurate, and timely nature of market analysis and strategic planning.

[0024] The following explanation, in conjunction with the accompanying drawings, uses several embodiments to illustrate the concepts. Figure 1 This is a flowchart illustrating a vehicle competition analysis method provided in an embodiment of this application, as shown below. Figure 1 As shown, the vehicle competition analysis method includes: S101, obtain vehicle replacement sales data within the target time period.

[0025] The sales data for replacement purchases should include at least: the category identifier and reference price of the preceding vehicle in at least one dimension, and the category identifier and reference price of the following vehicle in at least one dimension.

[0026] In this step, the "previous vehicle" refers to the original vehicle that is being replaced during the user's upgrade or replacement purchase, while the "replacement vehicle" refers to the target vehicle that the user is upgrading to.

[0027] Category identifiers are used to distinguish vehicle attributes across different dimensions. At least one dimension includes, but is not limited to, vehicle model, brand, energy type, body class, body type, manufacturer, and group. The same vehicle may correspond to different category identifiers across different dimensions. For example, a vehicle might be categorized as Brand A in the brand dimension, as pure electric in the energy type dimension, and as Class B in the body class dimension.

[0028] The reference price refers to the market reference price of the vehicle, which can be the manufacturer's suggested retail price, the average transaction price, or the median of the officially published price range, and is used for subsequent price distribution and similarity calculations.

[0029] The target time period can be set according to the analysis needs and dimensions. For example, the vehicle model dimension uses data from the past 6 months for semi-annual analysis, while the brand and company dimensions use data from the past 3 months for quarterly analysis. The selected time period must ensure that the data has a sufficient sample size and timeliness. It also supports a sliding window mechanism, which can be used to dynamically analyze competitive relationships by periodically sliding the time window.

[0030] One possible approach involves acquiring replacement and upgrade records covering all regions and vehicle models nationwide through an authoritative industry data platform, ensuring the comprehensiveness and authority of the data. First, data standardization is performed. Since the naming rules for vehicle models may differ in the original data (e.g., different power versions of the same vehicle series are listed as independent models), to avoid instability in statistical results due to overly fine data granularity, all subcategories of the same vehicle model need to be unified into a single model, standardizing the data statistical scope. Second, time windows are defined and key fields are extracted. Based on the set target time period, all transaction records within the time window are filtered from the data. Subsequently, key fields are extracted from the original data, including the preceding vehicle model, brand, manufacturer, group, body class, body type, energy type, and age; the following vehicle model, brand, manufacturer, group, body class, body type, energy type, replacement / upgrade volume, province, and city. Fields can be extracted or merged according to actual analytical needs. Extracting key fields clarifies the dimensions for analysis. In subsequent analysis, if we analyze the competitive relationship between vehicle models, the objects are (the preceding vehicle model and the following vehicle model); if we analyze the competitive relationship between brands, the objects are (the preceding vehicle brand and the following vehicle brand). Similarly, we can define different dimensions of analysis objects such as (the preceding vehicle company and the following vehicle company).

[0031] Furthermore, the original fields can be supplemented according to the analytical needs. If the original data lacks a specific analytical dimension field, the target field can be generated by mapping existing fields. For example, if the original data does not have a "Region" field, but regional analysis is required, the "Province" field can be added to each data record according to the preset geographical region division rules (such as mapping Hubei, Hunan, and Henan to "Central China Region"). Similarly, if it is necessary to analyze the city-level market, the "City" field can be mapped to generate line-level identifiers from "First-Tier Cities" to "Sixth-Tier Cities." This process ensures that the analysis can flexibly adapt to the research needs of various market segments.

[0032] Next, the data needs to be cleaned to remove interfering factors. First, each field of the data is validated. If a data entry contains field values ​​that do not conform to naming rules (e.g., incorrect brand name entry) or abnormal null values ​​in key fields (e.g., missing replacement / upgrade volume, vehicle model identification, etc.), that data entry is deleted and not included in subsequent statistical analysis. Simultaneously, low-priced models with extremely small data volumes in the replacement / upgrade market (e.g., models with less than 5 replacement / upgrade units per month) are removed. These niche models lack statistical significance, and their occasional replacement / upgrade behavior may interfere with the objectivity of the overall analysis results. For high-priced models, they are merged at the brand level to avoid insufficient sample size for a single high-priced model. Based on specific analytical needs, target market segment data is filtered using dimension fields. For example, to analyze the competition in the A-class gasoline vehicle market in Central China, the region field is used to filter for Central China, the vehicle class field for A-class, and the energy type field for gasoline. Only data meeting all filtering criteria is retained, forming a valid analytical dataset for the segmented market.

[0033] Through the above preprocessing and cleaning steps, high-quality sales data for replacement purchases are obtained. Each data point includes the category identifier and reference price of the preceding vehicle in at least one dimension, as well as the category identifier and reference price of the following vehicle in the corresponding dimension, providing a data foundation for subsequent calculation of relative preference index and price similarity.

[0034] S102, based on the category identifiers of the preceding and following vehicles in the target dimension from the sales data of replacement purchases, calculate the relative preference index between the preceding and following vehicles, and generate a relative preference matrix under the target dimension.

[0035] The target dimension is a pre-selected vehicle competition analysis dimension, which can be any one of the aforementioned at least one dimension. It can be determined according to the actual competition analysis needs. The selected target dimension matches the analysis target object defined in S101. If the analysis target object is the brand of the preceding and following vehicles in the brand dimension, then the corresponding target dimension is the brand dimension. If the analysis target object is the model of the preceding and following vehicles in the model dimension, then the corresponding target dimension is the model dimension, and so on. By selecting a single target dimension for analysis, the competitive relationship of vehicles under that dimension can be quantified. At the same time, different dimensions can be selected as target dimensions in sequence according to needs to complete multi-dimensional competition analysis.

[0036] In practical implementation, based on the replacement purchase sales data obtained from S101, and using the category identifier of the target dimension as a basis, a relative preference index is calculated between the preceding and following vehicles. This relative preference index quantifies the degree of replacement purchase preference of the user group corresponding to the preceding vehicle for a specific following vehicle. The index value directly reflects the attractiveness of the following vehicle to the user group to which the preceding vehicle belongs. Optionally, a preset ratio formula can be used to calculate this index in conjunction with the replacement purchase sales data under the target dimension, thereby achieving accurate quantification of replacement preferences under the segmented dimension. A relative preference matrix is ​​constructed based on the relative preference indices of all combinations of preceding and following vehicles. The rows and columns of the matrix correspond to the preceding and following vehicles under the target dimension, respectively, and the element values ​​in the matrix are the relative preference indices of the corresponding combinations of preceding and following vehicles, which can present the preference relationships between all preceding and following vehicles under the target dimension.

[0037] S103, Based on the reference price of the preceding vehicle in the sales data of replacement purchases, calculate the price similarity between all preceding vehicle objects and generate a preceding vehicle similarity matrix.

[0038] Price similarity is used to characterize the similarity of different preceding vehicles in terms of price. Its value directly reflects the similarity of the price positioning of different preceding vehicles. The higher the price similarity, the closer the price distribution characteristics of the two vehicles are, and the more likely they are to form a direct price competition relationship.

[0039] In practice, based on the sales data of replacement purchases obtained in S101, and using the reference prices of each preceding vehicle as a basis, price similarity is calculated among all preceding vehicle objects. Optionally, a preset price similarity formula can be used to calculate the price similarity between any two preceding vehicle objects, and each different combination of preceding vehicle objects corresponds to a unique price similarity result. (Precedence vehicle similarity matrix) It is a matrix constructed by integrating the price similarity results between all preceding objects. Its rows and columns correspond one-to-one with the preceding objects in the target dimension. The element values ​​in the matrix are the price similarity between preceding objects. This matrix can intuitively present the price similarity of all preceding objects in the target dimension.

[0040] S104: Based on the relative preference matrix and the preceding vehicle similarity matrix, obtain the competition intensity value between the target vehicle object and other vehicle objects in the target dimension.

[0041] Among them, the competition intensity value is the user's representation of the strength of competition between the target vehicle object and other vehicle objects in the target dimension.

[0042] In this step, based on the relative preference matrix generated in S102 and the preceding vehicle similarity matrix constructed in S103, the data from the preference dimension and the price dimension are fused through a preset comprehensive calculation logic to determine the competition intensity value between the target vehicle and all other vehicle objects. The calculation can be performed by using a kernel method. This method integrates the characteristics of user upgrade preferences represented in the relative preference matrix with the characteristics of price distribution similarity reflected in the similarity matrix of preceding vehicles. Through operations such as vector mapping, weight allocation, and standardization, the competition intensity value can be calculated. This ensures that the competition intensity value takes into account both user upgrade preferences and the price distribution similarity of preceding vehicles, thus ensuring the comprehensiveness and rationality of the competition intensity quantification.

[0043] Furthermore, after obtaining all competition intensity values, self-verification data and data distribution judgment can be performed to ensure the validity and discriminative power of the calculation results. Specifically, in the self-verification data judgment, the competition intensity value of a certain vehicle object to itself should always be 1. If other results occur, anomalies in the calculation process need to be checked. In the data distribution judgment, if the competition intensity values ​​are generally concentrated around 1, resulting in insufficient discriminative power, the preset coefficients in the previous vehicle similarity matrix generation process can be adjusted, or the off-diagonal elements of the previous vehicle similarity matrix can be further scaled to adjust the parameters. At the same time, if there are obvious outliers, it indicates that the preference data of the corresponding vehicle object may have statistical anomalies. The relative preference index of the corresponding vehicle object needs to be checked, the relevant preset parameters used in the calculation should be adjusted according to the actual business situation, and previous vehicle objects with insufficient sample size should be removed to ensure data sufficiency.

[0044] Furthermore, multi-dimensional market competition relationship analysis can be applied based on competition intensity values.

[0045] Specifically, in key competitor analysis, A higher value indicates a stronger competitive intensity between the two entities. This can be achieved by analyzing the competitive intensity value. Ranking is performed, and a list of its competitors is automatically generated. This allows for the identification of competitors for the target vehicle and the comparison of competitive relationships across different market segments. Since the calculation of competitive intensity is based on the target dimension, different market segments correspond to sub-dimensions within that target dimension (e.g., when the target dimension is vehicle type, market segments can be divided by energy type, such as pure electric vehicles, hybrid vehicles, etc.). Therefore, by filtering effective analytical data from different sub-dimensions, the competitive intensity value within the corresponding market segment can be calculated. This allows for comparison of differences in competitors and changes in competitive intensity rankings for the target vehicle across different market segments. For example, competitors for the target gasoline vehicle in first-tier city segments may differ from those in third- and fourth-tier city segments. Furthermore, it allows for comparison of the competitive intensity of the same competitor against the target vehicle in different city tiers, as well as comparison of the overall competitive level of each competitor in different city tiers compared to the entire market. Through cross-market segment competitive relationship comparisons, a more comprehensive understanding of the target vehicle's competitive situation can be achieved, enabling companies to develop more targeted regional strategies.

[0046] In the analysis of competitive relationship evolution, a sliding window mechanism is used to repeatedly execute the aforementioned steps on a monthly basis to obtain the competition intensity values ​​between target vehicle objects and other vehicle objects in multiple target time periods. This yields a multi-period competition intensity matrix (each period is on a monthly basis, denoted as ). For the target vehicle object that needs to be analyzed From each period Extract its competition intensity value with other vehicle objects, and filter out the top-ranked vehicles in each period. Each competitor object is constructed separately. The corresponding time series of competition intensity values, for example , , This allows us to reflect historical changes in the absolute intensity of competitive relationships; it also enables the construction of time series rankings of competitive intensity values, for example... , , This process reflects the historical fluctuations in the relative rankings of competitors. Subsequently, by applying algorithms to identify abnormal changes in time-series trends (such as the Mann-Kendall trend test combined with the Pettitt test), competitors whose competitive intensity values ​​or rankings have changed significantly are located. Finally, visualization methods are used to display the changing trends in the competitive relationship between these competitors and the target vehicle.

[0047] In constructing the competitive relationship graph, graph visualization algorithms (such as those in NetworkX) and drawing tools are used. Specifically, for each vehicle object, the competitors with the highest similarity are selected, and the competitive intensity value between them is retained as the edge weight in the graph. Then, based on the selected objects and edges, a network graph layout algorithm (such as Spring Layout) is used to build the competitive relationship graph. In the competitive relationship graph, nodes represent the analyzed objects, and the node size is positively correlated with the total transaction volume of that object in the upgrade and replacement market. Edges are used to reflect the competitive relationships between objects. When A is a top 10 competitor of B or B is a top 10 competitor of A, an edge is generated between their nodes. The thickness of the edge is determined by the competitive intensity value; the lower the competitive intensity value, the thinner the edge. Meanwhile, node distance can reflect competitive relationships. Generally, neighboring nodes have higher competitive intensity, but it needs to be judged comprehensively in combination with the existence of edges and the actual competitive intensity value. Furthermore, the automatic clustering function of the graph layout algorithm will aggregate similar brands / models into clusters. Objects in the same cluster are usually similar in price, positioning or user group. Objects in the center of the cluster are often key competitors in the segmented market. Outliers in the graph represent vehicle objects with relatively unique customer distribution and weaker competitive relationship with mainstream market participants.

[0048] All competitive intensity values ​​and analysis results can be stored in a structured database and subsequently imported into a BI (Business Intelligence) front-end system. Through time-series anomaly algorithms and graph visualization algorithms, analysis reports can be output, thereby providing data support for the marketing and product departments in strategy formulation and decision-making.

[0049] In this embodiment, massive amounts of real replacement and upgrade purchase data can be transformed into objective and quantifiable competitive intensity values. This allows automakers to clearly understand which products their own products directly compete with in specific dimensions, as well as the relative strength of the competition. Based on the competitive intensity values, companies can improve the scientific nature and accuracy of their decisions when formulating market strategies, allocating marketing resources, and evaluating product positioning, effectively responding to market competition and significantly enhancing the scientific nature, accuracy, and timeliness of market analysis and strategic planning.

[0050] In the above Figure 1 Based on the corresponding embodiments, in order to more clearly demonstrate the process of generating the relative preference matrix, this application also provides a possible implementation of generating the relative preference matrix in a vehicle competition analysis method. Figure 2 This is a schematic diagram illustrating the process of generating a relative preference matrix in a vehicle competition analysis method provided in this application embodiment. Figure 2As shown, in S102 above, based on the category identifiers of the preceding and following vehicles in the target dimension from the sales data of replacement purchases, a relative preference index is calculated between the preceding and following vehicles, generating a relative preference matrix in the target dimension, which includes: S210, based on the sales data of replacement purchases, groups the vehicle pairs to obtain multiple replacement purchase vehicle pairs.

[0051] Specifically, transaction records in the replacement and upgrade sales data are grouped based on the category identifier of the target dimension. Specifically, replacement and upgrade records with the same preceding vehicle category identifier and the same following vehicle category identifier are grouped together to form a replacement and upgrade vehicle pair. Each replacement and upgrade vehicle pair corresponds to a combination of preceding and following vehicles under the target dimension. For example, when the target dimension is brand, all replacement and upgrade transaction records for preceding vehicle brand A and following vehicle brand B will be grouped into one replacement and upgrade vehicle pair.

[0052] S220, based on the category identifier of the preceding vehicle in each vehicle replacement / replacement pair in the target dimension, calculate the relative preference index between the preceding and following vehicles in the vehicle replacement / replacement pair, and obtain the relative preference index of the vehicle replacement / replacement pair.

[0053] Based on each vehicle pair obtained for replacement or upgrade purchases, and combined with its corresponding sales data, a relative preference index is calculated between the preceding and following vehicles, using the category identifier of the target dimension as a basis. Specifically, the calculation process uses the category identifier of the preceding vehicle (e.g., brand A) and the category identifier of the following vehicle (e.g., brand B) as data query conditions. Relevant transaction records are filtered and summarized from the sales data for replacement and upgrade purchases, thereby obtaining the overall replacement data related to the category identifier of the preceding vehicle and the overall inflow data related to the category identifier of the following vehicle. On this basis, a preset ratio formula is used to obtain the relative preference index corresponding to each vehicle pair for replacement or upgrade purchases. The higher the value of the relative preference index, the stronger the attraction of the following vehicle to the user group of the preceding vehicle, thus quantifying the replacement preference of a specific preceding and following vehicle combination under a segmented dimension.

[0054] S230 generates a relative preference matrix under the target dimension based on the relative preference index of multiple vehicle replacement and upgrade pairs.

[0055] After calculating the relative preference index for all vehicle replacement / exchange pairs, a relative preference matrix is ​​constructed under the target dimension. Specifically, with all preceding vehicles under the target dimension as rows of the matrix and all following vehicles as columns, the relative preference index for each vehicle replacement / exchange pair is filled into the corresponding positions in the matrix according to the correspondence between preceding and following vehicles.

[0056] For combinations of preceding and following vehicles with no actual replacement or upgrade transaction records under the target dimension, the corresponding matrix element values ​​can be set to preset default values ​​(such as 0) to ensure the integrity of the matrix. The resulting relative preference matrix can intuitively present the replacement preference relationships between all preceding and following vehicles under the target dimension, providing structured data support for subsequent calculations of vehicle competition intensity.

[0057] In this embodiment, by grouping replacement purchase data into replacement purchase vehicle pairs and calculating the relative preference index for each replacement purchase vehicle pair, a relative preference matrix is ​​finally constructed. This transforms massive discrete transaction records into structured and quantifiable preference relationships, significantly improving the automation and objectivity of extracting competitiveness from the data.

[0058] In the above Figure 2 Based on the corresponding embodiments, in order to more clearly demonstrate the process of generating the relative preference matrix, this application also provides a possible implementation of generating the relative preference matrix in a vehicle competition analysis method. Figure 3 This is a flowchart illustrating the generation of a relative preference matrix in another vehicle competition analysis method provided in this application embodiment. For example... Figure 3 As shown in S220 above, based on the category identifier of the preceding vehicle in each vehicle replacement / exchange pair in the target dimension, and the category identifier of the preceding vehicle in the target dimension, the relative preference index between the preceding and following vehicles in the vehicle replacement / exchange pair is calculated. The relative preference index of the vehicle replacement / exchange pair includes: S310: Based on the category identifier of the preceding vehicle in the target dimension and the category identifier of the preceding vehicle in the target dimension, obtain the total number of people who have replaced the preceding vehicle and the total number of people who have replaced the preceding vehicle.

[0059] Specifically, for all preceding vehicles in the target dimension With the object behind The combination, count all previous vehicle objects This refers to all trade-in transactions for the initial vehicle, i.e., the previous vehicle. Trade-in to any vehicle within the target dimension (in, The total transaction amount (representing any subsequent vehicle object) is used to obtain the transaction amount of the preceding vehicle object. The total number of car owners who have engaged in trade-in purchases is denoted as . This indicates that the preceding vehicle will be... Trade in your car to every possible replacement Add up all the quantities here. Iterate through all existing following vehicle objects (such as all brands or all models) under the target dimension, which covers the preceding vehicle objects. This involves understanding the replacement needs of the user group and summarizing all previous competitors within the target category. (in, (Representing any previous vehicle) traded in to a subsequent vehicle. The transaction records show that the vehicle was traded in to obtain the replacement vehicle. The total quantity is denoted as This means including all possible preceding vehicles. Trade-in to the next vehicle Add up all the quantities here. Iterate through all existing preceding objects in the target dimension.

[0060] S320: Based on the total number of people who replaced their vehicles in the preceding vehicle category and the total number of vehicles replaced from the preceding vehicle category to the following vehicle category in the replacement vehicle category, calculate the conditional probability of replacing the vehicle with the following vehicle category.

[0061] Specifically, for all preceding vehicles in the target dimension With the object behind The combination, based on the preceding vehicle object Total number of people who exchanged Next, count all objects belonging to the preceding vehicle. Users who trade in their vehicles to other vehicles The transaction records are recorded as the total exchange quantity. This allows for the calculation of the vehicle that can be traded in. conditional probability See formula (1) to calculate conditional probability. : Formula (1) in Indicates that the object in front is Under the premise that the user trades in their vehicle for another one The probability. For example, if the object in front... Total number of people who exchanged Among them, the replacement of the vehicle to the target quantity ,but That is, the vehicle in front 20% of users' trade-in needs are directed towards their next car. Conditional probability eliminates the influence of the overall market size and focuses on the preceding events. The proportion of users who choose to trade in their vehicles within a user group can accurately reflect the preference of a specific preceding user for a target following vehicle.

[0062] S330: Calculate the marginal probability of acquiring a replacement vehicle based on the total number of replacement vehicles and the total number of replacement vehicles acquired by the replacement vehicle group.

[0063] Specifically, for all preceding vehicles in the target dimension With the object behind The combination, based on the trade-in target vehicle. Total number Next, we will count the total number of vehicles purchased by customers who are upgrading or replacing their vehicles, which represents the total transaction volume of the entire upgrade and replacement market. , ,in, Represents any preceding vehicle object. Represents any following vehicle object. This represents all transaction records from any preceding vehicle to any subsequent vehicle. Then, the transaction to the subsequent vehicle is calculated according to formula (2). marginal probability : Formula (2) in This indicates that in the overall trade-in market, any trade-in transaction is a replacement vehicle. The probability. For example, if the total market transaction volume... The object behind the car Total number of trade-ins ,but That is, the object following the vehicle Occupying 8% of the replacement purchase market share, the marginal probability reflects the target of the next vehicle purchase. Its market appeal is not limited by a specific group of previous users.

[0064] S340. Based on the conditional probability and marginal probability of the replacement vehicle, calculate the relative preference index between the front vehicle and the rear vehicle in the replacement vehicle pair, and obtain the relative preference index of the replacement vehicle pair.

[0065] Specifically, the conditional probability calculated based on S320 Marginal probability calculated with S330 Solve for the preceding vehicle object With the object behind Relative preference index between The calculation formula is shown in formula (3): Formula (3) when At that time, it indicates the object in front. User's view of the following vehicle The trade-in preference is significantly higher than the market average, and the target of the subsequent vehicle purchase is... For the vehicle in front Its user base has a differentiated advantage in terms of appeal; when At that time, the degree of preference was consistent with the market average, and the subsequent target... For the vehicle in front The appeal to users conforms to general market laws; when At that time, the degree of preference was lower than the market average, and the subsequent target was... For the vehicle in front The user base is relatively less attractive.

[0066] Using the above calculation method, all preceding vehicle objects in the target dimension can be calculated. With the object behind The relative preference index of the combination ultimately forms a comprehensive index covering all vehicle replacement and upgrade pairs. The relative preference matrix of size (where For the number of vehicles in front, (Number of vehicles following).

[0067] In this embodiment, statistical trade-in data is used to derive a relative preference index by combining conditional probability and marginal probability. This accurately reflects the differentiated preferences of the preceding vehicle user for the following vehicle, enabling a fair comparison of preference intensity among different vehicle objects under a unified scale. The resulting relative preference index can objectively and accurately quantify the attractiveness of the following vehicle object to a specific preceding vehicle user, providing interpretable data for subsequent competition intensity analysis.

[0068] In the above Figure 2 Based on the corresponding embodiments, to more clearly demonstrate the process of optimizing the relative preference matrix, this application also provides a possible implementation of optimizing the relative preference matrix in a vehicle competition analysis method. Optionally, based on the above S210-S230, the method further includes: S410 optimizes the relative preference index of each vehicle replacement / replacement pair based on the number of vehicle replacement / replacement pairs and the preset relative preference index.

[0069] In practical applications, due to the long-tail distribution of sales data for replacement and upgrade vehicles, the transaction sample size of some preceding or following vehicle pairs within the target time period may be extremely limited. For these replacement and upgrade vehicle pairs with insufficient sample size, the directly calculated observed RPI values ​​are often unstable and prone to abnormally high or low values ​​due to accidental factors, which is detrimental to subsequent competition intensity analysis. Therefore, in this embodiment, a pre-defined heuristic Bayesian smoothing method is used to optimize the relative preference index of each replacement and upgrade vehicle pair.

[0070] Specifically, for each vehicle purchase / replacement pair (previous vehicle object) With the object behind Its optimized relative preference index The calculation is shown in formula (4): Formula (4) in, To determine the sample size corresponding to the current vehicle replacement / replacement purchase pair, the preceding vehicle in that pair can be selected. Trade-in to the next vehicle Total number of trade-ins Or the vehicle in front Total number of people who exchanged ; The relative preference index before optimization, calculated in the aforementioned steps; It is the prior relative preference index for vehicle replacement purchases. When the sample data is insufficient, direct observation is unreliable and prior knowledge needs to be introduced for constraint. For example, when the analysis object is the vehicle model dimension, the overall RPI value of the brand to which the vehicle model belongs can be used as its prior value. When the analysis object is the brand dimension, the overall average RPI level of the company to which the brand belongs can be used as its prior value. By introducing statistics of higher level and larger sample as prior values, a reasonable benchmark can be provided for lower level objects in the absence of direct data support. It is the smoothing coefficient, also known as the confidence parameter, which represents the sample size required for prior values ​​to have reference value. The larger the value is set, the more the smoothing result tends to favor the prior value, and the stronger the suppression effect on outlier observations. The smaller the value setting, the more the smoothing result tends to resemble the observed value, and the higher the degree of preservation of the original data. In practical applications, The value can be configured based on business experience. A preferred implementation is to use the median of the sample size distribution of all vehicle replacement / exchange pairs as the benchmark. This allows for smoothing of most regular sample sizes and applies stronger shrinkage to small samples at the tail.

[0071] After completing all vehicle replacement and upgrade purchases... After calculation, for A comprehensive observation and verification of the distribution was conducted: if the vast majority All concentrated in Nearby indicates the current setting. The value is too large, causing the Bayesian smoothing to be too strong, resulting in a distortion that deviates from the actual exchange preferences. In this case, readjustment is necessary. Setting the value, reducing Substitute the value again The relative preference index of each vehicle purchase / replacement pair before optimization is calculated and then re-smoothed until... The distribution can accurately and objectively reflect the actual replacement preference relationship between the preceding and following vehicles under the target dimension, and optimize the relative preference index of each replacement vehicle pair.

[0072] In this embodiment, by optimizing the relative preference index, the problem of an abnormally large relative preference index caused by small samples is effectively solved. By combining prior values ​​with an adjustable smoothing coefficient and verifying the optimization parameters according to the distribution of the relative preference index, the optimized relative preference index is made to better fit the actual exchange scenario, improve data accuracy, and provide a reliable quantitative basis for subsequent analysis.

[0073] In the above Figure 1 Based on the corresponding embodiments, in order to more clearly demonstrate the process of generating the preceding vehicle similarity matrix, this application also provides a possible implementation method for generating the preceding vehicle similarity matrix in a vehicle competition analysis method. Figure 4 This is a schematic diagram illustrating the process of generating a preceding vehicle similarity matrix in a vehicle competition analysis method provided in an embodiment of this application. Figure 4 As shown in S103 above, based on the reference price of the preceding vehicle in the replacement purchase sales data, the price similarity between all preceding vehicle objects is calculated, generating a preceding vehicle similarity matrix, which includes: S510: Based on the reference price of the previous vehicle in the sales data of replacement purchases, obtain the number of vehicles of the previous vehicle in each preset price range.

[0074] Specifically, based on sales data from replacement and upgrade purchases, the reference prices of all previous vehicle models are extracted, and the price range is divided according to preset price segmentation rules. A continuous price range can be defined, for example, the price range can be divided into several intervals such as 0-100,000, 100,000-200,000, 200,000-500,000, 500,000-1,000,000, and above 1,000,000. Then, the reference prices of each preceding vehicle are categorized, and each preceding vehicle is recorded individually. Falling within each preset price range The number of vehicles within the range is used to form the distribution data of the number of vehicles in different price ranges for each preceding vehicle.

[0075] S520: Based on the total number of preceding vehicles and the number of vehicles in each preset price range, obtain the probability distribution of preceding vehicles relative to each preset price range.

[0076] Specifically, for each preceding vehicle, the calculation is performed by using the number of its vehicles in each preset price range as the numerator and the total number of its vehicles as the denominator. In each preset price range The percentage of vehicles within the specified range is normalized to obtain the probability distribution of each preset price range for the preceding vehicle, denoted as [missing information]. ,in Representing the The first person in front was in the first The probability values ​​for a preset price range, where the sum of all probability values ​​is 1, represent the probability of the preceding vehicle. The probability distribution form along the price dimension; similarly, the... The price probability distribution of the preceding vehicles is denoted as . .

[0077] S530: Calculate the price distribution distance between different preceding objects based on the probability distribution of different preceding objects.

[0078] First, construct a cost matrix and extract the original price formation list for each preset price range. ( For the first The original prices for each price range are used to obtain the square root transformation for each original price point. ,(in , For the first (the price value after the square root transformation of each price range), and then according to the formula... Construct a cost matrix, where Indicates from the first The price range moves to the first Unit cost for each price range ( , Then, the Wasserstein Distance method was used to calculate the distance between any two preceding vehicles. and Price distribution distance between That is, the Earth Mover's Distance (EMD), which is calculated using formula (5): Formula (5) in, and The first and Price probability distribution of the preceding vehicle object and The cumulative distribution function, in the context of discrete price intervals, is transformed into a summation calculation over each price segment position to obtain the price distribution distance between each pair of preceding vehicles. .

[0079] S540, calculate the price similarity between different preceding objects among all preceding objects based on the price distribution distance between different preceding objects.

[0080] Based on any two preceding vehicle objects obtained above and Price distribution distance between By using a preset similarity conversion rule, distance is mapped to price similarity value, thereby obtaining the price similarity between different preceding objects among all preceding objects.

[0081] Specifically, for any two preceding vehicles... and If the price distribution of the two is far apart The larger the value, the more significant the difference in price distribution characteristics, and the higher the corresponding price similarity. The smaller it is; if The smaller the value, the closer the price distribution characteristics are, and the higher the corresponding price similarity. The larger the similarity, the more consistent the similarity and distance can be in terms of monotonicity through preset similarity conversion rules, while also normalizing the value range. This facilitates the subsequent construction of a similarity matrix.

[0082] In actual calculations, all preceding vehicle object pairs are traversed along the target dimension. The price distribution distance between different preceding objects Substituting the preset similarity conversion rules, the corresponding price similarity is obtained one by one. The final generated price similarity dataset for all preceding vehicle pairs shows that the closer the value is to 1, the more similar the price distribution of the two preceding vehicles is, and the higher the correlation in competing for the same segment of the customer base; the closer the value is to 0, the greater the difference in the price distribution of the two preceding vehicles is, and the weaker the competition in the price dimension.

[0083] S550 generates a similarity matrix of preceding vehicles based on the price similarity between multiple groups of different preceding vehicle objects.

[0084] Collect the price similarity values ​​between all pairs of preceding objects. Using the preceding vehicle objects as the rows and columns of a matrix, the corresponding similarity values ​​are filled into the corresponding positions of the matrix according to the pairing relationship of the preceding vehicle objects, thus constructing a matrix... Symmetric similarity matrix of vehicles in front ( (This represents the total number of preceding vehicles). Meanwhile, to avoid excessively high similarity values ​​for preceding vehicles leading to subsequent competition strength values ​​mostly approaching 1 and thus losing their discriminative meaning, the off-diagonal elements of the preceding vehicle similarity matrix are scaled to reduce the influence of similarity between non-preceding vehicles, ultimately resulting in a preceding vehicle similarity matrix with effective discriminative power.

[0085] In this embodiment, by quantifying the price distribution of preceding vehicles, combining it with the calculation of the price distribution distance between preceding vehicles, and then converting it into price similarity, a forward similarity matrix is ​​finally constructed, realizing the quantitative representation of competitive association in the price dimension and improving the accuracy of vehicle competition analysis.

[0086] In the above Figure 4 Based on the corresponding embodiments, to more clearly demonstrate the process of calculating price similarity, this application also provides a possible implementation of calculating price similarity in a vehicle competition analysis method. Optionally, in the above-described S540, calculating the price similarity between different preceding vehicle objects based on the price distribution distance between different preceding vehicle objects includes: S610 uses a preset attenuation coefficient to map the price distribution distance between different preceding objects, thereby obtaining the price similarity between different preceding objects.

[0087] Based on the above S530, any two preceding vehicle objects are calculated. and Price distribution distance between The distance is mapped to a price similarity value between 0 and 1 using a preset attenuation coefficient. For details, please refer to formula (6): Formula (6) in, For the vehicle in front and The similarity of price distribution between them For the vehicle in front and The price distribution distance between them This is a preset adjustable attenuation coefficient used to control the intensity of the distance's influence on similarity. The larger the value, the more identical. The lower the price similarity obtained by mapping, the more sensitive it is to the price distribution differences between preceding objects, and the more finely it can distinguish preceding objects with similar distributions. The smaller the value, the lower the sensitivity of price similarity to changes in distance. Even if there are some differences in distribution, the price similarity may still remain at a high level. Prices can be flexibly configured and optimized according to actual business scenarios, data distribution characteristics, and analysis granularity.

[0088] In one possible implementation, further, after calculating the competition intensity value, if the data distribution analysis reveals that the vast majority of competition intensity values ​​are abnormally close to 1, it indicates that the competition intensity index lacks discriminative power and cannot effectively identify the differences in competition strength between different objects. In this case, it can be determined as the attenuation coefficient. If the setting is too small, the differences between the off-diagonal elements in the similarity matrix of the preceding vehicle will not be obvious enough. In this case, it needs to be increased appropriately. The values ​​are then substituted back into formula (6) for mapping calculation until the distribution of competition intensity values ​​has a reasonable degree of differentiation.

[0089] In this embodiment, the price distribution distance is mapped to the price similarity value through an exponential decay function, realizing the transformation of price distribution differences into similarities. The adjustable decay coefficient can adjust the sensitivity of price distribution differences, effectively avoiding the problem of insufficient discrimination of similarity indicators, and making the output price similarity more in line with business analysis needs.

[0090] In the above Figure 1 Based on the corresponding embodiments, in order to more clearly demonstrate the process of obtaining the competition intensity value, this application also provides a possible implementation method for obtaining the competition intensity value in a vehicle competition analysis method. Figure 5 This is a schematic diagram illustrating the process of obtaining competition intensity values ​​in a vehicle competition analysis method provided in an embodiment of this application. Figure 5 As shown, in S104 above, based on the relative preference matrix and the preceding vehicle similarity matrix, the competition intensity value between the target vehicle object and other vehicle objects in the target dimension is obtained, including: S710, based on the relative preference matrix, obtain the first relative preference coefficient of the target vehicle object relative to the corresponding preceding vehicle object, and the second relative preference coefficients of other vehicle objects relative to the corresponding preceding vehicle object.

[0091] Based on the relative preference matrix under the target dimension, and using the target vehicle object (following vehicle object) as the retrieval criterion, the relative preference index of the target vehicle object relative to all preceding vehicle objects is extracted and used as the first relative preference coefficient, denoted as . (in For any preceding vehicle, (Target vehicle object); simultaneously extract the relative preference index of all other vehicle objects (the following vehicles to be analyzed) relative to all preceding vehicle objects, and use it as the second relative preference coefficient. (in For any preceding vehicle, To remove (Any other vehicle object besides the one mentioned above).

[0092] S720: Obtain the first vector of the target vehicle object based on the first relative preference coefficient of the target vehicle object and the market share of the corresponding preceding vehicle object.

[0093] Specifically, first obtain the objects of each preceding vehicle. market share Perform a square root transformation on it to obtain Then, the first relative preference coefficient of the target vehicle object is... With the corresponding preceding vehicle Multiplication, with the element values ​​of the corresponding dimension of the first vector, the first vector... Construct reference formula (6): Formula (6) in, The index represents the total number of preceding vehicles. Corresponding to the 1st, 2nd, and 3rd respectively The person ahead of me.

[0094] S730 obtains the second vector of other vehicle objects based on the second relative preference coefficient of other vehicle objects and the market share of the corresponding preceding vehicle object.

[0095] Specifically, using the same vector construction rules as S720, the first step is to obtain the objects of each preceding vehicle. market share Perform a square root transformation on it to obtain Then, the second relative preference coefficients of other vehicle objects are... With the corresponding preceding vehicle Multiplication, with the element values ​​of the corresponding dimension of the second vector, the second vector... Construct reference formula (7): Formula (7) All other vehicle objects complete the second vector construction according to this rule, thereby obtaining the second vector of all other vehicle objects.

[0096] S740: Based on the first vector of the target vehicle object, the second vector of other vehicle objects, and the similarity matrix of the preceding vehicle, obtain the competition intensity value between the target vehicle object and other vehicle objects in the target dimension.

[0097] Specifically, based on the first vector of the target vehicle object The second vector of other vehicle objects and the similarity matrix of the preceding vehicle The weighted similarity of two vectors is calculated to obtain the competition intensity value between the target vehicle and other vehicle objects under the target dimension. During the calculation, the similarity matrix of the preceding vehicle is used as the weight to fuse the first and second vectors, achieving a correlation between them in the dimensions of preceding vehicle user preferences and price distribution similarity. A higher competition intensity value indicates a stronger competition between the target vehicle and other vehicle objects for the same preceding vehicle user group. This process is repeated for all other vehicle objects, calculating the competition intensity value for each one, ultimately yielding the competition intensity value between the target vehicle and all other vehicle objects.

[0098] Furthermore, by traversing all pairs of following vehicle objects under the target dimension, and based on the competition intensity values ​​between the target vehicle object and all other vehicle objects, as well as the competition intensity values ​​between all pairs of following vehicle objects, a system is constructed. Competition intensity matrix of size ( (Number of vehicles following).

[0099] In this embodiment, a feature vector is constructed based on the relative preference matrix and the market share of the preceding vehicle, and the competition intensity value is calculated by integrating the preceding vehicle similarity matrix. This realizes the correlation between replacement preference and price dimension competition, providing accurate and comprehensive quantitative basis for vehicle competition analysis and improving the scientificity and accuracy of competition intensity assessment.

[0100] In the above Figure 5 Based on the corresponding embodiments, in order to more clearly demonstrate the process of obtaining the competition intensity value, this application also provides a possible implementation method for obtaining the competition intensity value in a vehicle competition analysis method. Figure 6 This is a schematic diagram illustrating the process of obtaining the competition intensity value in another vehicle competition analysis method provided in this application embodiment. Figure 6 As shown, in S740 above, based on the first vector of the target vehicle object, the second vectors of other vehicle objects, and the similarity matrix of the preceding vehicle, the competition intensity value between the target vehicle object and other vehicle objects in the target dimension includes: S810: Perform a dot product of the first vector of the target vehicle object, the second vector of other vehicle objects, and the similarity matrix of the preceding vehicle to obtain the dot product vector of the target vehicle object and other vehicle objects.

[0101] Specifically, based on the first vector of the target vehicle object The second vector of other vehicle objects Combined with the similarity matrix of the preceding vehicle The generalized dot product is used to calculate the dot product vector between the target vehicle object and other vehicle objects. The similarity matrix of the preceding vehicle contains... Represents two preceding objects and Price similarity between them, satisfying , Due to the symmetry properties, the generalized dot product can be calculated using formula (8): Formula (8) in, The total number of vehicles in front. For target vehicle object The first vector in the second Component values ​​on the dimension of the preceding vehicle object; For other vehicle objects The second vector in the th Component values ​​on the dimension of the preceding vehicle object; The first vehicle in the similarity matrix of the preceding vehicle The first vehicle object and the first Price similarity between preceding vehicles.

[0102] Generalized dot product through the similarity matrix of the preceding vehicles The object in front and The similarity between them is introduced into the calculation, so that when hour, At this point, the dot product vector of the target vehicle object and other vehicle objects is ;when At that time, if the vehicle in front is and If the price distributions of two vehicles are highly similar, the dot product vector of the target vehicle and other vehicles will also increase. This means that even if two following vehicles do not correspond to the same preceding vehicle, as long as the price similarity between the two preceding vehicles is high, this competitive relationship will still be reflected in the final competitive intensity value.

[0103] S820: Based on the similarity matrix of the preceding vehicles, calculate the vector norms of the first vector of the target vehicle object and the second vectors of the other vehicle objects respectively, and obtain the first vector norm and the second vector norm.

[0104] Specifically, the front vehicle similarity matrix As weights, respectively, for the first vector of the target vehicle object The second vector of other vehicle objects Calculate the generalized vector norm to obtain the first vector norm. Second vector norm The formula for calculating the generalized vector norm is given in formula (9): Formula (9) in, The total number of vehicles in front. For vector identifiers, take The first vector corresponding to the target vehicle object ,Pick The second vector corresponding to other vehicle objects ; For vectors In the The component values ​​of each preceding vehicle object dimension For vectors In the Component values ​​on the dimension of the preceding vehicle object; The first vehicle in the similarity matrix of the preceding vehicle The first vehicle object and the first Price similarity between preceding vehicles.

[0105] This calculation incorporates the price similarity between preceding vehicles as a weight into the vector magnitude calculation, achieving standardized vector processing and providing a unified quantitative basis for subsequent calculations.

[0106] S830: Based on the dot product vector of the target vehicle object and other vehicle objects, as well as the first vector norm and the second vector norm, the competition intensity value between the target vehicle object and other vehicle objects in the target dimension is obtained.

[0107] Specifically, the dot product vector of the target vehicle object and other vehicle objects obtained based on S810. And the first vector norm obtained from S820 Second vector norm The kernel cosine similarity algorithm is used to calculate the competition intensity value between the target vehicle object and other vehicle objects. (Value), the calculation formula is shown in formula (10): Formula (10) in That is, the target vehicle object in the target dimension. Other vehicle objects The competition intensity value between them. The competition intensity value is essentially a value derived from the preceding vehicle similarity matrix. In the defined kernel space, the cosine of the angle between two rear vehicle object vectors has a range of values. , The higher the value, the greater the similarity between the two following vehicle objects under the dual backgrounds of user preferences and price, meaning the more intense the competition between them for the same user group. By traversing all pairs of following vehicle objects under the target dimension, a competition intensity matrix can be generated. Each value in the matrix represents the degree of competitive association between the corresponding two following vehicle objects under the target dimension.

[0108] In this embodiment, the competition intensity value is obtained by combining the generalized dot product with the similarity matrix of the preceding vehicle and normalizing it with the generalized vector norm. Then, the competition intensity value is obtained through the kernel cosine similarity. This realizes the quantification of the competitive relationship between user preferences and price dimension, so that the output competition intensity value can reflect the actual competitive relationship between the following vehicles.

[0109] The following describes a vehicle competition analysis device and electronic device provided in this application, which are used to perform the application. The specific implementation process and technical effects are described above and will not be repeated below.

[0110] Figure 7 This is a schematic diagram of a vehicle competition analysis device provided in an embodiment of this application, as shown below. Figure 7 As shown, the vehicle competition analysis device includes: Module 1000 is used to acquire vehicle replacement and upgrade sales data within a target time period. This sales data includes at least: the category identifier and reference price of the preceding vehicle in at least one dimension, and the category identifier and reference price of the following vehicle in at least one dimension.

[0111] The calculation module 2000 is used to calculate the relative preference index between the preceding and following vehicles based on the category identifiers of the preceding and following vehicles in the target dimension from the sales data of replacement purchases, and to generate a relative preference matrix in the target dimension; to calculate the price similarity between all preceding vehicles based on the reference prices of the preceding vehicles in the sales data of replacement purchases, and to generate a preceding vehicle similarity matrix; and to obtain the competition intensity value between the target vehicle and other vehicle objects in the target dimension based on the relative preference matrix and the preceding vehicle similarity matrix.

[0112] Optionally, the calculation module 2000 is specifically used to group vehicle pairs according to the sales data of replacement purchases, to obtain multiple replacement purchase vehicle pairs; to calculate the relative preference index between the preceding and following vehicles in each replacement purchase vehicle pair according to the category identifier of the preceding vehicle in the target dimension and the category identifier of the preceding vehicle in the target dimension, to obtain the relative preference index of the replacement purchase vehicle pair; and to generate a relative preference matrix under the target dimension based on the relative preference indices of multiple replacement purchase vehicle pairs.

[0113] Optionally, the acquisition module 1000 is specifically used to acquire the total number of people who have traded in the preceding vehicle and the total number of vehicles that have traded in to the following vehicle, based on the category identifier of the preceding vehicle in the target dimension and the category identifier of the preceding vehicle in the target dimension.

[0114] Optionally, the calculation module 2000 is specifically used to calculate the conditional probability of switching to the next vehicle based on the total number of people switching to the previous vehicle and the total number of vehicles switching from the previous vehicle to the next vehicle in the vehicle pair; to calculate the marginal probability of switching to the next vehicle based on the total number of vehicles switching to the next vehicle and the total number of vehicles switching to the next vehicle in the vehicle pair; and to calculate the relative preference index between the previous and next vehicles in the vehicle pair based on the conditional probability and the marginal probability of switching to the next vehicle, thus obtaining the relative preference index of the vehicle pair.

[0115] Optionally, the calculation module 2000 is also used to optimize the relative preference index of each vehicle replacement pair based on the number of replacement vehicle pairs and the preset relative preference index.

[0116] Optionally, module 1000 is specifically used to obtain the number of vehicles of the previous vehicle in each preset price range based on the reference price of the previous vehicle in the sales data of the replacement purchase.

[0117] Optionally, the calculation module 2000 is specifically used to obtain the probability distribution of the preceding vehicle objects relative to each preset price range based on the total number of preceding vehicle objects and the number of vehicles in each preset price range; calculate the price distribution distance between different preceding vehicle objects based on the probability distribution of different preceding vehicle objects; calculate the price similarity between different preceding vehicle objects among all preceding vehicle objects based on the price distribution distance between different preceding vehicle objects; and generate a preceding vehicle similarity matrix based on the price similarity between multiple sets of different preceding vehicle objects.

[0118] Optionally, the calculation module 2000 is specifically used to map the price distribution distance between different preceding objects using a preset attenuation coefficient, so as to obtain the price similarity between different preceding objects.

[0119] Optionally, the calculation module 2000 is specifically used to obtain, based on the relative preference matrix, a first relative preference coefficient of the target vehicle object relative to its corresponding preceding vehicle object, and a second relative preference coefficient of other vehicle objects relative to their corresponding preceding vehicle objects; to obtain a first vector of the target vehicle object based on the first relative preference coefficient of the target vehicle object and the market share of its corresponding preceding vehicle object; to obtain a second vector of other vehicle objects based on the second relative preference coefficient of other vehicle objects and the market share of their corresponding preceding vehicle objects; and to obtain the competition intensity value between the target vehicle object and other vehicle objects in the target dimension based on the first vector of the target vehicle object, the second vector of other vehicle objects, and the preceding vehicle similarity matrix.

[0120] Optionally, the calculation module 2000 is specifically used to perform a dot product of the first vector of the target vehicle object, the second vector of other vehicle objects, and the similarity matrix of the preceding vehicle to obtain the dot product vector of the target vehicle object and other vehicle objects; based on the similarity matrix of the preceding vehicle, to calculate the vector norms of the first vector of the target vehicle object and the second vector of other vehicle objects respectively to obtain the first vector norm and the second vector norm; and based on the dot product vector of the target vehicle object and other vehicle objects, as well as the first vector norm and the second vector norm, to obtain the competition intensity value between the target vehicle object and other vehicle objects in the target dimension.

[0121] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more digital signal processors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).

[0122] Figure 8 This is a schematic diagram of an electronic device provided in an embodiment of this application. The device may be a computing device or a server with computing processing capabilities.

[0123] The electronic device 10 includes a processor 11, a storage medium 12, and a bus 13. The storage medium 12 stores program instructions executable by the processor 11. When the electronic device 10 is executed, the processor 11 communicates with the storage medium 12 via the bus 13, and the processor 11 executes the program instructions to perform the above-described method embodiment. The specific implementation and technical effects are similar and will not be described in detail here.

[0124] Optionally, this application also provides a program product, such as a computer-readable storage medium, including a program that, when executed by a processor, performs the above-described method embodiments.

[0125] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0126] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0127] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in a combination of hardware and software functional units.

[0128] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0129] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A vehicle competition analysis method, characterized in that, The method includes: Obtain vehicle replacement and purchase sales data within a target time period; wherein, the replacement and purchase sales data includes at least: the category identifier of the preceding vehicle in at least one dimension and its reference price, and the category identifier of the following vehicle in at least one dimension and its reference price; Based on the category identifier of the preceding vehicle in the target dimension and the category identifier of the following vehicle in the target dimension in the sales data of the replacement purchase, calculate the relative preference index between the preceding vehicle and the following vehicle, and generate the relative preference matrix under the target dimension. Based on the reference price of the preceding vehicle in the sales data of the replacement purchase, calculate the price similarity between all the preceding vehicle objects and generate a preceding vehicle similarity matrix; Based on the relative preference matrix and the preceding vehicle similarity matrix, the competition intensity value between the target vehicle object and other vehicle objects in the target dimension is obtained.

2. The method according to claim 1, characterized in that, The step of calculating the relative preference index between the preceding vehicle and the following vehicle in the target dimension based on the category identifier of the preceding vehicle and the following vehicle in the target dimension from the sales data of the replacement purchase includes: Based on the sales data of the replacement purchase, vehicle pairs are grouped to obtain multiple replacement purchase vehicle pairs; Based on the category identifier of the preceding vehicle in the target dimension and the category identifier of the preceding vehicle in the target dimension in each vehicle replacement / replacement pair, calculate the relative preference index between the preceding vehicle and the following vehicle in the vehicle replacement / replacement pair to obtain the relative preference index of the vehicle replacement / replacement pair. A relative preference matrix under the target dimension is generated based on the relative preference indices of multiple vehicle replacement / replacement purchase pairs.

3. The method according to claim 2, characterized in that, The step of calculating the relative preference index between the preceding and following vehicles in each vehicle replacement / exchange pair based on the category identifier of the preceding vehicle in the target dimension and the category identifier of the preceding vehicle in the target dimension, to obtain the relative preference index of the vehicle replacement / exchange pair, includes: Based on the category identifier of the preceding vehicle in the target dimension among multiple vehicle replacement and upgrade pairs, and the category identifier of the preceding vehicle in the target dimension, obtain the total number of people who have upgraded the preceding vehicle and the total number of vehicles that have upgraded to the following vehicle. Based on the total number of people who traded in the preceding vehicle and the total number of vehicles traded in from the preceding vehicle to the following vehicle in the vehicle pair, calculate the conditional probability of trading in to the following vehicle. The marginal probability of acquiring a vehicle is calculated based on the total number of vehicles acquired through trade-in and the total number of vehicles acquired through trade-in in the vehicle pair. Based on the conditional probability and marginal probability of acquiring the next vehicle, the relative preference index between the previous vehicle and the next vehicle in the vehicle replacement pair is calculated to obtain the relative preference index of the vehicle replacement pair.

4. The method according to claim 2, characterized in that, The method further includes: Based on the number of the vehicle replacement / replacement pairs and the preset relative preference index, the relative preference index of each vehicle replacement / replacement pair is optimized.

5. The method according to claim 1, characterized in that, The step of calculating the price similarity among all preceding vehicles based on the reference price of the preceding vehicle in the replacement purchase sales data, and generating a preceding vehicle similarity matrix, includes: Based on the reference price of the preceding vehicle in the sales data of the replacement purchase, obtain the number of vehicles of the preceding vehicle in each preset price range; Based on the total number of preceding vehicles and the number of vehicles in each preset price range, obtain the probability distribution of the preceding vehicles relative to each preset price range. Calculate the price distribution distance between different preceding vehicles based on the probability distribution of the different preceding vehicles; Calculate the price similarity between different preceding objects based on the price distribution distance between different preceding objects; The preceding vehicle similarity matrix is ​​generated based on the price similarity between multiple sets of different preceding vehicle objects.

6. The method according to claim 5, characterized in that, The step of calculating the price similarity between different preceding vehicles based on the price distribution distance between different preceding vehicles includes: By using a preset attenuation coefficient, the price distribution distance between different preceding vehicle objects is mapped to obtain the price similarity between different preceding vehicle objects.

7. The method according to claim 1, characterized in that, The step of obtaining the competition intensity value between the target vehicle object and other vehicle objects in the target dimension based on the relative preference matrix and the preceding vehicle similarity matrix includes: Based on the relative preference matrix, obtain the first relative preference coefficient of the target vehicle object relative to the corresponding preceding vehicle object, and the second relative preference coefficients of other vehicle objects relative to the corresponding preceding vehicle object; Based on the first relative preference coefficient of the target vehicle object and the market share of the corresponding preceding vehicle object, obtain the first vector of the target vehicle object; Based on the second relative preference coefficient of the other vehicle objects and the market share of the corresponding preceding vehicle object, obtain the second vector of the other vehicle objects; Based on the first vector of the target vehicle object, the second vector of the other vehicle objects, and the similarity matrix of the preceding vehicle, the competition intensity value between the target vehicle object and the other vehicle objects in the target dimension is obtained.

8. The method according to claim 7, characterized in that, The step of obtaining the competition intensity value between the target vehicle object and other vehicle objects in the target dimension based on the first vector of the target vehicle object, the second vector of the other vehicle objects, and the preceding vehicle similarity matrix includes: The dot product vector of the target vehicle object and the other vehicle objects is obtained by performing a dot product on the first vector of the target vehicle object, the second vector of the other vehicle objects, and the similarity matrix of the preceding vehicle. Based on the preceding vehicle similarity matrix, the vector norms of the first vector of the target vehicle object and the second vector of the other vehicle objects are calculated respectively to obtain the first vector norm and the second vector norm; Based on the dot product vector of the target vehicle object and the other vehicle objects, as well as the first vector norm and the second vector norm, the competition intensity value between the target vehicle object and the other vehicle objects in the target dimension is obtained.

9. A vehicle competition analysis device, characterized in that, The device includes: The acquisition module is used to acquire vehicle replacement and purchase sales data within a target time period; wherein, the replacement and purchase sales data includes at least: the category identifier of the preceding vehicle in at least one dimension and its reference price, and the category identifier of the following vehicle in at least one dimension and its reference price. The calculation module is used to calculate the relative preference index between the preceding vehicle and the following vehicle in the target dimension based on the category identifier of the preceding vehicle in the replacement purchase sales data and the category identifier of the following vehicle in the target dimension, and generate a relative preference matrix in the target dimension; calculate the price similarity between all preceding vehicle objects based on the reference price of the preceding vehicle in the replacement purchase sales data, and generate a preceding vehicle similarity matrix; and obtain the competition intensity value between the target vehicle object and other vehicle objects in the target dimension based on the relative preference matrix and the preceding vehicle similarity matrix.

10. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores program instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus, and the processor executes the program instructions to perform the steps of the vehicle competition analysis method as described in any one of claims 1 to 8.

Images