Consumption Probability Prediction Method, Device, Medium and System Based on Hesitancy Degree Analysis

Abstract

The present invention discloses a consumption probability prediction method, device, medium and system based on hesitation degree analysis, belonging to the technical field of consumption probability prediction, including: collecting data, including consumers' online activity data and offline navigation software usage data, and determining the spatial location of the sales point and the commodity brand information from the collected data; preprocessing the collected data to form sales point dimension data; mining the correlation and mutual influence between the sales point dimension data, and performing consumption probability prediction based on hesitation degree analysis. Based on the analysis of the change in consumption ability, the present invention considers the key factors in the entire behavior process of consumers, can fully reveal the correlation between consumers' travel and related positioning information and online activities, and improves the accuracy and practicability of consumption prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of consumption probability prediction, and more specifically, to a consumption probability prediction method, device, medium and system based on hesitation degree analysis. Background Art

[0002] With the popularization of smart phones and the wide application of navigation software, consumers are increasingly inclined to use navigation software to plan routes to offline stores. In the field of large commodities, there may be a close connection between consumers' online activities (such as searching, browsing, consulting, etc.) and offline store visits during the process of using navigation software to the store. However, there is currently no effective method in the market to comprehensively and systematically analyze this correlation.

[0003] In the existing consumption probability prediction solutions, there are the following problems: due to the lack of attention to the change in consumption ability and the lack of attention to the key factors in the entire behavior process of consumers, the existing analysis methods have deficiencies in revealing the correlation between travel and related location information and online activities, resulting in poor analysis accuracy and low practicality, and it is difficult to provide enterprises with more refined market insight capabilities. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a consumption probability prediction method, device, medium and system based on hesitation degree analysis. Based on the analysis of the change in consumption ability, considering the key factors in the entire behavior process of consumers, it can fully reveal the correlation between consumers' travel and related location information and online activities, and improve the accuracy and practicality of consumption prediction.

[0005] The purpose of the present invention is achieved through the following solutions:

[0006] A consumption probability prediction method based on hesitation degree analysis includes the following steps:

[0007] Collect data, including consumers' online activity data and offline navigation software usage data, and determine the spatial location of the sales point and the commodity brand information from the collected data;

[0008] Based on the operation of a program by a computer processor, preprocess the collected data to form sales point dimension data;

[0009] Based on the operation of a program by a computer processor, mine the correlation and mutual influence between the sales point dimension data, and perform consumption probability prediction based on hesitation degree analysis.

[0010] Further, determining the spatial location of the point of sale from the collected data specifically includes sub-steps: retrieving the collected data through the Internet, cross-checking and comparing it with the names publicly available on the official website and the recommended names to determine the accurate name of the point of sale and the longitude and latitude coordinates, finally delineating the target location, adjusting the range, and then cross-checking through third-party data to ultimately determine the spatial location of the point of sale.

[0011] Further, the consumer online activity data includes search records, browsing history, and consultation content; the offline navigation software usage data includes departure location, arrival time, route selection, entry and exit time, stay duration, portrait information, arrival method, search for the corresponding point of sale, viewing, route planning, and navigation travel data; the commodity brand information includes independent sub-brands; the adjusted range specifically includes delineating the capture range of the visiting behavior of the point of sale destination, and the range includes the store exhibition hall and the external public area of the store.

[0012] Further, preprocessing the collected data to form point-of-sale dimension data specifically includes sub-steps: performing preprocessing operations such as cleaning, deduplication, and time alignment on the collected data to ensure the consistency and comparability of the data, and finally using the data obtained by associating the destination data with the brand, category, and surrounding environment as the point-of-sale dimension data.

[0013] Further, preprocessing the collected data specifically includes preprocessing the basic information data of the point of sale. In the preprocessing of the basic information data of the point of sale, it specifically includes sub-steps:

[0014] a), using the permutation and combination of brand, name, and address to perform keyword searches within the scope of the store's city using the spatial search interface. Each spatial query of each combination will obtain multiple return results, and summarize and count the number of occurrences of all the IDs obtained from the permutation and combination.

[0015] b), sorting the summarized IDs according to the number of occurrences, selecting the number of occurrences of the top first set value of the IDs for comparison. If the number of occurrences of the first one ≥ the second set value, it is considered that the ID of this store in the spatial data of the professional mapping company is this ID. Otherwise, perform inverse geocoding on the store's address, compare the distance with the coordinates corresponding to the ID, and perform a comprehensive score based on the distance between the inverse geocoding and the coordinates corresponding to the ID and the number of occurrences of the ID. Finally, select the one with the highest score as the ID of the store.

[0016] c), checking the ID determined through distance comparison on the maps of each mapping company and making corrections according to the checking results.

[0017] d), comparing with the basic information data of the point of sale collected by the vertical media platform to determine its relevance, and then complementing the information missing on the official website, which includes the sold categories and models.

[0018] Further, the preprocessing of the collected data specifically includes arrival data preprocessing, and in the arrival data preprocessing, it specifically includes sub-steps:

[0019] The arrival data includes the stay duration and the entry and exit times; within the capture range, the visit records with a stay duration less than the set value and the entry and exit times outside the store business hours are all excluded as abnormal data. If the stay duration is a null value, it is filled with the mode;

[0020] Dimensionality reduction according to the arrival time: According to the passenger flow curve of the commodity sales area, the arrival data is dimensionally reduced into entering the store during peak hours and entering the store during non-peak hours;

[0021] Dimensionality reduction according to the arrival date: According to the characteristics of arriving at the store on weekdays and holidays, the store arrival data is divided into arriving at the store on weekdays and arriving at the store on non-weekdays;

[0022] Dimensionality reduction according to the arrival judgment accuracy: According to the way of arriving at the store, it is distinguished between accurate arrival at the store and inaccurate arrival at the store; the distinguishing methods include: if the navigation destination is the same as the store entering the capture fence, it is regarded as accurate arrival at the store, otherwise it is regarded as inaccurate arrival at the store.

[0023] Further, comparing with the basic information data of the sales points collected by the vertical media platform to determine its relevance, and then complementing the information missing on the official website, specifically including sub-steps:

[0024] 1), Normalize and standardize the name and address;

[0025] 2), Calculate and compare the similarity of the address and name in the same province and city, and distinguish absolute coincidence, very high coincidence, need for spot check, and need for manual intervention judgment according to the score.

[0026] Further, mining the relevance and mutual influence between the sales point dimension data, and predicting the consumption probability based on the hesitation degree analysis, specifically including sub-steps:

[0027] Construct a consumer hesitation degree model, specifically including destination retrieval hesitation degree, destination planning hesitation degree, destination arrival hesitation degree, and hesitation degree of arriving at the destination and then leaving, and use the analysis of the consumer hesitation degree to predict the consumption probability.

[0028] Further, the method for determining the destination retrieval hesitation degree specifically includes the following sub-steps:

[0029] a1: Obtain the user's true destination from the preprocessed data;

[0030] a2: De-correlate the destination and compare the categories and brands;

[0031] a3: Analyze the retrieval behavior from the sales point dimension data; specifically, it includes the following situations:

[0032] i. If the search behavior is for multiple brands and multiple stores, the current consumption probability is judged to be particularly hesitant, and multiple brands and multiple stores are further distinguished as follows:

[0033] If the multiple brands are from completely different categories: The hesitancy level is the highest, which means that consumers are in the early exploration stage, their needs are unclear, they are conducting research, and they are in the overall planning stage;

[0034] If the multiple brands are multiple brands of similar categories: the degree of hesitation is lower than the highest, the consumer's search behavior is concentrated, the comparison is between multiple brands, there is a clear understanding of the product category, there is uncertainty in brand selection, and in-depth brand and product comparisons are being conducted to find the best choice;

[0035] The determination of the destination search hesitation is used to confirm that consumers have transitioned from online preparation to offline travel, and some brands and categories have been preliminarily selected for on-site inspection;

[0036] ii. If the search behavior is to search for only a single brand, the current consumption probability is judged to be generally hesitant, and the degree of hesitation is further subdivided according to the search frequency as follows:

[0037] If the search frequency is lower than the first set frequency, it is considered a low-frequency search, which is used to indicate that the consumer has a certain degree of knowledge of the brand and is only making a preliminary understanding or confirming the information, and is still considering it;

[0038] If the search frequency is lower than the second set frequency, it is judged as a medium frequency search, which is used to indicate that the consumer is interested in the product and is seriously considering purchasing it, but has not made up his mind and is comparing different models or waiting for the right time;

[0039] If the search frequency is lower than the third set frequency, it is judged as a high-frequency search, which is used to indicate that the consumer is very interested in the brand, but is obviously hesitant and is struggling with the specific model, price or whether to buy it;

[0040] The search frequency can indicate that consumers have already decided on a brand, or that they are loyal to the brand, are hesitant about the category and price, and want to compare the categories and prices of multiple stores of the same brand. The more hesitant a consumer is, the closer he is to making a decision at the current stage.

[0041] iii. If the search behavior is for multiple stores of a single brand, the current consumption probability is judged to be generally hesitant, and the inclination to choose a store is judged based on the distance between the residence and workplace and the store as follows:

[0042] If the distances between the place of residence and the place of work and the store are less than the first set distance, it is judged that the selection preference is the highest, and the convenience of arriving at the store is the main consideration factor;

[0043] If the distances between the place of residence and the place of work and the store are less than the second set distance, it is judged that the selection preference is medium, weighing between the convenience of arriving at the store and other factors, and looking for better options or specific services;

[0044] If the distances between the place of residence and the place of work and the store are less than the third set distance, it is judged that the selection preference is medium and the lowest, but if the store is selected as the subsequent travel destination, it is of great significance. Such consumers must have reasons for choosing to give up convenience;

[0045] By combining the judgment of the distances between the place of residence and the place of work and the store, it can be characterized that consumers already have a clear brand choice in mind, or plan to compare prices, and are hesitating about prices at this time, planning to compare prices among multiple stores in the product category of the same brand, or still considering geographical location, service quality factors;

[0046] iv. If the retrieval behavior is a clear single store, it is judged that the current consumption probability is not hesitant. At this time, consumers already have a clear target brand and store in mind. Such retrieval indicates that consumers have very clear purposes. Although the purposes are clear, there may still be hesitation or changes at the last moment;

[0047] a4: Based on the analysis of retrieval behavior, form a judgment conclusion in the decision-making stage of retrieval: Divide the hesitation degree of destination retrieval into particularly hesitant, generally hesitant, and not hesitant, and sort according to the retrieval hesitation degree. The lower the hesitation degree, the closer to the decision on the final brand category and planned travel;

[0048] According to the analysis of the hesitation degree of retrieval behavior, the higher the retrieval hesitation degree, the earlier the stage of offline on-site visit to the store, the less clear the focus on the target store, and even there is no target; the lower the hesitation degree, the clearer the focus of consumers, and the greater the possibility of subsequent in-store consumption; formulate different marketing and promotional strategies according to different hesitation degrees to promote the reduction of consumers' hesitation degree and influence the decision-making direction of consumers. The consumer group with retrieval hesitation is more forward in the decision-making process. It can be considered to provide detailed brand and product comparison information in marketing promotion activities, highlight the unique advantages of its own brand, provide more experience and trial services, and launch price matching or other preferential policies.

[0049] Furthermore, the determination method of the destination planning hesitation degree specifically includes the following sub-steps:

[0050] b1: Analyze the travel planning behavior according to the data of the sales point dimension; specifically, it includes the following situations:

[0051] i. If the destination travel planning behavior is in the multi-brand and multi-store scenario, it is judged as particularly hesitant, which is used to characterize that the consumer hesitates among a limited number of brand categories, but the willingness to visit stores has been very strong;

[0052] ii. If the destination travel planning behavior is single-brand and multi-store, it is judged as generally hesitant, which is used to characterize that the consumer has already selected a brand, and has even carried out actual visits to multiple brands, initially determined the category, and the planning behavior is in the comparison and screening of distance, travel mode, and store visit order, or has entered the price comparison stage;

[0053] iii. If the destination travel planning behavior is single-brand and single-store, it is judged as not hesitant, which is used to characterize that the consumer's destination is very clear, and the planning behavior is to confirm the travel route or determine the public transportation transfer method, and further analyze as follows in combination with the planned travel mode:

[0054] If traveling by car, such consumers already have a car, and the hesitation factor is the road condition, and this purchase is a trade-in or an additional purchase;

[0055] If traveling by bus, such consumers may be making their first purchase, and the hesitation factors may be the number of transfers and the total travel time;

[0056] If traveling on foot or by bike, such customers tend to choose stores at a short distance, but usually in the early stage of visiting stores. Combining with subsequent planning data, if there is a planned visit to a store of the same brand at a long distance, it indicates that the decision-making stage is progressing continuously and is closer to the final decision;

[0057] b2: Based on the analysis of the planning behavior, a judgment conclusion is formed in the decision-making stage of the plan: the destination planning hesitation degree is divided into particularly hesitant, generally hesitant, and not hesitant. Sort according to the planning hesitation degree, and the lower the hesitation degree, the closer to the final store selection decision; Optionally, different targeted marketing means are used according to different hesitation degrees and the consumer's target brand and store to guide the consumer's decision-making.

[0058] Furthermore, the specific method for determining the destination arrival hesitation degree specifically includes the following sub-steps:

[0059] c1: Analyze the destination arrival hesitation degree based on the sales point dimension data; specifically, the following situations are included:

[0060] i. If the consumer arrives only at the planned store, it is judged as not hesitant, which is used to characterize that the consumer has a clear goal, knows what to buy, has conducted sufficient research, the store visit behavior highly conforms to the expected plan, the final decision is approaching, and the purchase intention is strong. Set the consumer probability expectation to high;

[0061] ii. If the consumer not only arrives at the planned store but also visits other unplanned stores, it is judged as general hesitation, which is used to characterize that the consumer has not fully made up their mind. Besides the planned store, they also want to make more comparisons. The decision is affected by new information after on-site inspection, approaching the final decision, but still making the final comparison. Set their consumption probability expectation as medium;

[0062] iii. If the consumer does not arrive at the planned store but arrives at other stores instead, it is judged as special hesitation, which is used to characterize that the consumer only has a preliminary interest, or the original plan has changed due to various reasons, or the preselected brand has been changed. They are in the early stage of the decision-making process, and their needs or preferences may change significantly. Set their consumption probability expectation as low;

[0063] iv. If the consumer does not arrive at all, that is, they do not arrive at any store completely, it is judged as special hesitation, which is used to characterize that it may be just a temporary intention to view store information, without a strong willingness to purchase, just a preliminary exploration, or restricted by objective factors such as time and distance. The decision-making process is in the earliest stage and has not entered the actual purchase consideration stage yet. Set their consumption probability as the lowest;

[0064] c2: Hesitation degree analysis based on whether the destination is reached, forming a judgment conclusion in the decision-making stage of destination arrival: The destination arrival hesitation degree is divided into special hesitation, general hesitation, and no hesitation. Sort according to the destination hesitation degree, and the lower the hesitation degree, the closer to the final purchase decision.

[0065] In the above solution, from the time dimension analysis: The longer the time interval from planning to arrival, usually the higher the hesitation degree; The arrival time on weekdays vs weekends may reflect different purchase urgencies, and the urgency on weekdays is higher; From the space dimension analysis: The closer the distance between the place of residence / work and the target store, the higher the convenience, and the hesitation degree may be lower; On the contrary, if the original plan is still adhered to, it may indicate a strong purchase intention; From the visit frequency analysis: Multiple visits to the store may indicate strong interest and deeper hesitation; From the number of store brands visited analysis: Visiting stores of the same brand may indicate a high degree of loyalty to the brand or the brand has been selected, and it is very likely to be comparing prices and is already very close to the final decision; Visiting stores of different brands indicates that the intended model may have been determined and the final comparison and decision are being made;

[0066] In a further solution, match marketing strategies according to different analysis results: For the group with no hesitation: Strengthen the purchase confidence and provide a convenient transaction process; For the group with general hesitation: Provide detailed product comparison information and highlight unique advantages; For the group with special hesitation: Increase brand exposure, provide more experience opportunities, and provide more selection and comparison tools.

[0067] Furthermore, the method for determining the hesitation degree when the destination arrives and then leaves specifically includes the following sub-steps:

[0068] d1: The hesitation degree of arriving at and leaving the destination analyzed from the point-of-sale dimension data; specifically, it includes the following situations: Analyzing from the duration of consumers' stay in the store:

[0069] i. If the consumer stays in the store for more than the first set duration, and conducts in-depth understanding of the product, but finally does not purchase, it is judged as general hesitation, which is used to characterize a strong interest in the product, the decision-making stage is close to purchase, but there are still final concerns, and the possibility of their return to the store is set as high;

[0070] ii. If the consumer stays in the store within the set range, and views the product, but leaves without in-depth understanding, it is judged as general hesitation, which is used to characterize an interest in the product, but not strong enough, the decision-making stage is still in the consideration stage, and more information or time is needed, and the possibility of their return to the store is set as medium;

[0071] iii. If the consumer stays in the store for less than the second set duration, and leaves after a quick browse, it is judged as special hesitation, which is used to characterize little interest in the product, the decision-making stage is in need of re-evaluating the needs or considering other options, and the possibility of their return to the store is set as low;

[0072] d2: Forming a judgment conclusion based on the analysis of the duration of consumers' stay in the store: Classify the hesitation degree of leaving after arriving into special hesitation, general hesitation, and no hesitation, and sort according to the destination hesitation degree. The lower the hesitation degree, the higher the conversion rate.

[0073] Furthermore, in the method for determining the hesitation degree of arriving at and leaving the destination, the following sub-steps are also included:

[0074] Combining the analysis of consumers' behavior patterns after leaving the store, specifically including the following situations:

[0075] iv. If the consumer returns to the store again after leaving the store and there is no visit to other stores in between, it is judged as no hesitation;

[0076] v. If the consumer visits 1 - 3 other stores after leaving the store, it is judged as general hesitation;

[0077] vi. If the consumer visits more than 3 other stores or does not visit any store after leaving the store, it is judged as special hesitation.

[0078] Furthermore, in the method for determining the hesitation degree of arriving at and leaving the destination, the following sub-steps are also included:

[0079] Combining the analysis of the number of times consumers return to the store, specifically including the following situations:

[0080] i. If the consumer returns to the store 1 - 2 times after leaving the store, it is judged as no hesitation;

[0081] ii. If the customer returns to the store 3 times after leaving the store, it is judged as general hesitation;

[0082] iii. If the customer returns to the store more than 3 times after leaving the store: special hesitation.

[0083] Furthermore, after predicting the consumption probability by analyzing the hesitation degree of consumers, the method further includes the steps of formulating different marketing and promotion strategies according to different hesitation degrees to promote the reduction of consumers' hesitation degree and influence the decision-making direction of consumers, specifically including the following situations:

[0084] Provide discounts or time-limited promotions for the non-hesitating group to promote quick decision-making, strengthen the product advantages, provide exclusive customer service, simplify the purchase process, and provide convenient payment options;

[0085] Provide detailed product comparison information for the general hesitation group, arrange in-depth experience activities, and provide flexible purchase plans;

[0086] For the particularly hesitant group, increase brand exposure, provide basic education content, provide stress-free consultation services, invite them to participate in brand activities, and enhance emotional connection.

[0087] A consumption probability prediction device based on hesitation degree analysis, including:

[0088] A data acquisition module for collecting data, and the data includes consumers' online activity data and offline navigation software usage data, and determining the selling point spatial location and product brand information from the collected data;

[0089] A data preprocessing module for preprocessing the collected data based on a program run by a computer processor to form selling point dimension data;

[0090] A data mining module for mining the correlation and mutual influence between the selling point dimension data based on a program run by a computer processor, and predicting the consumption probability based on hesitation degree analysis.

[0091] A computer-readable storage medium stores a computer program, and the computer program, when loaded by a processor, executes the method described in any one of the above.

[0092] A consumption probability prediction system based on hesitation degree analysis includes the consumption probability prediction device based on hesitation degree analysis described above.

[0093] The beneficial effects of the present invention include:

[0094] (1) The present invention provides a new technical solution for predicting consumption probability, which can determine the relationship between Internet location data and human consumption behavior (especially for large consumer goods, such as cars), conduct pre-behavior analysis, such as search, planning, and browsing data of relevant APPs, as well as residence analysis, including duration, frequency, and distance, and correlation analysis, such as data of 4S stores of the same brand, the number of 4S stores of the same grade, and license plate registration or used car transaction data, etc. Based on the above analysis, sales volume prediction, store management, and precise advertising placement are carried out at the macro or micro level.

[0095] (2) The present invention can help merchants gain insights into the real potential customers among the store visitors, study the customer decision-making path, identify and analyze competitors, and assist in optimizing the sales funnel and adjusting sales strategies through the sales volume prediction data.

[0096] (3) Through the analysis of the relationship between people and places, the present invention can exclude non-potential purchase customers, thereby more accurately determining the number of "consumers" arriving at the store, improving the accurate insight into the number of people entering the store, and predicting future sales volume based on the number of consumers. BRIEF DESCRIPTION OF THE DRAWINGS

[0097] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0098] Figure 1 It is the data flow of potential customer mining in the embodiment of the present invention;

[0099] Figure 2 It is the flowchart of data preprocessing of the basic information of the sales point in the embodiment of the present invention;

[0100] Figure 3 It is the hesitation degree of destination retrieval in the embodiment of the present invention;

[0101] Figure 4 It is the hesitation degree of destination planning in the embodiment of the present invention;

[0102] Figure 5 It is the hesitation degree of destination arrival in the embodiment of the present invention;

[0103] Figure 6 It is the hesitation degree of arriving at the destination and then leaving in the embodiment of the present invention;

[0104] Figure 7 It is the structural block diagram of the device in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0105] All features disclosed in all embodiments in this specification, or steps in all methods or processes implicitly disclosed, except for mutually exclusive features and / or steps, can be combined and / or extended and replaced in any manner.

[0106] In a preferred embodiment, it particularly relates to the fields of Internet location data analysis, consumer behavior science, and location-based service technologies. Specifically, an analysis technical solution for the correlation between the travel behaviors of bulk commodity consumers and their online activities is proposed, and a consumption probability prediction model solution based on hesitation degree analysis is proposed. The present invention provides a consumption probability prediction method based on hesitation degree analysis, and the overall process is as Figure 1 shown. In the input, the upstream data includes an arrival data table, a portrait data table, an appendix of portrait data (residence), and an arrival mode data table. The self-built data includes a whitelist data table, a blacklist data table, and a store dimension data table. In the intermediate results, a logical blacklist data table and a potential customer data table can be formed, and a business output table is formed in the output. More specifically, in one implementation manner, the following steps are included:

[0107] Step S1, data collection

[0108] In the concept of the data collection step of the present invention, consumer online activity data (such as search records, browsing histories, consultation contents, etc.) and offline navigation software usage data (such as departure locations, arrival times, route selections, etc.) are integrated. For the retrieval data for navigation purposes, through Internet retrieval, cross-checking and cross-comparing are carried out using the official names and recommended names publicly available on the official website to determine the accurate names and longitude and latitude coordinates, and finally the target locations are circled and the ranges are adjusted, and then cross-checked through third-party data to finally determine the spatial locations of the sales points.

[0109] Furthermore, in the specific implementation manner of data collection, it includes but is not limited to the following ways:

[0110] (1) Use Internet retrieval to collect the basic information of sales points of each brand;

[0111] (2) Collect the mainstream brands (including independent sub-brands) of this bulk commodity currently on the domestic market;

[0112] (3) Use data collection means such as crawlers and customized interfaces to collect the basic information data of sales points (including sales point names, provinces, cities, detailed addresses, map longitude and latitude coordinates), categories, etc. from the official website and mainstream vertical media platforms;

[0113] (4) Determine the unique spatial correspondence relationship of the sales points on the map;

[0114] (5) Circle the capture range of the visit behaviors of the sales point destinations, and the range includes the store exhibition hall and the external public areas of the store (such as the parking lot area, the parking lot entrance and exit, etc.);

[0115] (6) The third party provides the visit information (entry and exit time and length of stay) of the people entering the visit capture range, portrait information, arrival method, search, viewing, route planning and navigation travel data of the corresponding sales points.

[0116] Step S2: Data preprocessing

[0117] In the data preprocessing step of the present invention, the collected data is cleaned, deduplicated, time aligned and other preprocessing operations are performed to ensure the consistency and comparability of the data, and finally the destination data is associated with the brand, category and surrounding environment to form data as point of sale dimension data.

[0118] Furthermore, in the specific implementation of data preprocessing, Figure 2 As shown, it includes the preprocessing of basic information data of the point of sale and the preprocessing of arrival data.

[0119] (1) Preprocessing of basic information data of sales points, including the following sub-steps:

[0120] a) Use the spatial search interface to search for keywords in the city where the store is located using the permutations and combinations of brand, name, and address. Each spatial query of each combination usually returns multiple results. The IDs obtained from all permutations and combinations are aggregated and counted for their occurrence times.

[0121] b) Sort the aggregated IDs by the number of occurrences, select the number of occurrences of the IDs ranked before the set value (e.g., 2) for comparison, if the number of occurrences of the first place is ≥ the set value (e.g., 7 times), it is considered that the ID of this store in the spatial data of the professional map vendor is this ID, otherwise, the store address is reverse geocoded and the distance is compared with the coordinates corresponding to the ID, and a comprehensive score is given based on the distance between the reverse geocoding and the coordinates corresponding to the ID and the number of ID occurrences, and finally the one with the highest score is selected as the store ID;

[0122] c) Check the ID determined by distance comparison on the maps of each map provider and make corrections based on the check results;

[0123] d) Compare the basic information data of the sales points collected by the vertical media platform to determine their relevance, and then complete the information such as sales categories and models that may be missing on the official website. Specifically, it includes:

[0124] 1) Use CPCA to normalize and standardize names and addresses;

[0125] 2) In the same province or city, the address and name are compared using similarity calculations. A score of 1 indicates an absolute match, a score of 0.8 or above indicates a very good match, a score of 0.7-0.8 requires spot checks, and a score of less than 7 requires manual intervention.

[0126] 3) Through the above steps, the association relationship between the stores on the vertical domain platform and the stores announced on the official website can be determined.

[0127] (2) Reaching data preprocessing, specifically including sub-steps:

[0128] a) The reaching data includes arrival time, stay duration, and departure time. Since the acquisition result of the user's location information is comprehensively affected by various factors such as the positioning usage method of the app, the usage habits of the user, and the network situation, the stay duration may be very short or even missing when collected within the capture range. It is necessary to process this part of abnormal data; for those that are too short (with a stay duration less than 5 minutes) and arrival / departure times outside the store business hours, the visit records are excluded as abnormal data. For null values of the stay duration, the mode is used for filling;

[0129] b) Dimensionality reduction according to the arrival time: According to the usual passenger flow curve in the commodity store, usually 10:30 - 12:00 and 13:30 - 16:00 are the peak periods for customers to enter the store. Therefore, the reaching data is dimensionally reduced into entering the store during peak periods and entering the store during non-peak periods;

[0130] c) Dimensionality reduction according to the arrival date: Similarly, according to the characteristics of arriving at the store on weekdays and holidays, the arriving store data is divided into arriving at the store on weekdays and arriving at the store on non-working days;

[0131] d) Dimensionality reduction according to the accuracy of arrival judgment: According to the way of arriving at the store, if the navigation destination is the same as the store entering the capture fence, it is regarded as arriving at the store accurately, and other ways are regarded as not arriving at the store accurately.

[0132] Step S3, Hesitation degree analysis

[0133] Using statistical and machine learning algorithms (such as time series analysis, path analysis, clustering analysis, etc.) to deeply analyze the preprocessed online and offline data, and establish a consumption hesitation degree model to explore the correlation and mutual influence between them.

[0134] Specifically, the determination method of the consumption hesitation degree model includes sub-steps: Constructing hesitation degree indicators: destination retrieval hesitation degree, destination planning hesitation degree, hesitation degree of arriving at the destination, and hesitation degree of arriving at the destination and then leaving.

[0135] A: As Figure 3 shown, the determination method of the destination retrieval hesitation degree is as follows:

[0136] a1: Obtain the user's true destination from the preprocessed data;

[0137] a2: De-associate the destination and compare categories and brands;

[0138] a3: Analyze the search behavior based on the point of sale dimension data, including:

[0139] i. Multiple brands and multiple stores: especially hesitant;

[0140] This may also mean that consumers are conducting extensive market research or are not clear about their needs. Further segmentation:

[0141] Multiple brands in completely different categories: extremely hesitant, in the early stages of exploration, with unclear consumer needs, conducting extensive research, and may still be in the overall planning stage;

[0142] Multiple brands in similar categories: The degree of hesitation is high. Consumers’ search behavior is relatively concentrated, but they are still comparing multiple brands. They have a clear understanding of the product category, but are uncertain about brand selection. They are conducting in-depth brand and product comparisons to find the best choice.

[0143] Consumers have moved from online preparation to offline travel, and have preliminarily selected some brands and categories and plan to check them out in person.

[0144] ii. Search only a single brand: generally hesitant;

[0145] Consumers have already determined the brand in their mind, which may also mean that they have a high degree of loyalty to the brand, are hesitant about the category and price, and hope to compare the categories and prices among multiple stores of the same brand. The degree of hesitation of such hesitant consumers can be further subdivided by combining the search frequency:

[0146] Low-frequency searches (1-2 times / week): Consumers have a certain level of knowledge about the brand and may only be doing preliminary research or confirming information. They are still considering the brand and are more likely to make decisions about going out to visit stores.

[0147] Medium frequency searches (3-4 times / week): The consumer is interested in the product and is seriously considering purchasing it, but has not yet made up his mind. He may be comparing different models or waiting for the right time.

[0148] High frequency search (5 times or more per week): Consumers are very interested in the brand, but are obviously hesitant and may be struggling with the specific model, price or whether to buy it;

[0149] Combined with the search frequency, the more hesitant and conflicted consumers are, the closer they are to making a decision at the current stage.

[0150] iii. Single brand with multiple stores: generally hesitant;

[0151] Consumers already have a clear brand choice in mind and may plan to compare prices. At this time, they are hesitant about the price and plan to compare the prices of the same brand across multiple stores. They may also be considering factors such as geographical location and service quality.

[0152] Such hesitant consumers combine the distances between their place of residence and workplace and the store to judge their tendency to choose a store:

[0153] Short distance (within 5 km): High tendency to choose, and the convenience of getting to the store is the main consideration factor;

[0154] Medium distance (5 - 15 km): Medium tendency to choose. Consumers weigh between convenience and other factors and may be looking for better options or specific services;

[0155] Long distance (more than 15 km): Low tendency to choose. However, if the store is chosen as the subsequent travel destination, it is of great significance. Such consumers must have special reasons for giving up convenience and may be looking for unique products, better prices or services.

[0156] iv. Definite single store: Without hesitation;

[0157] Consumers already have a definite target brand and store in mind. Such searches indicate that consumers have very clear purposes. Although the purposes are clear, there may still be hesitation or changes at the last moment.

[0158] According to the analysis of the hesitation degree of the search behavior, the higher the search hesitation degree, the earlier the stage of offline on-site visit to the store, the less clear the attention to the target store, and there may even be no target; the lower the hesitation degree, the clearer the attention target of the consumer, and the greater the possibility of subsequent store visits.

[0159] a4: Judgment conclusion in the decision-making stage of the search: The search hesitation degree is divided into particularly hesitant, generally hesitant, and not hesitant. Sort according to the search hesitation degree. The lower the hesitation degree, the closer to the final decision on the brand category and planned trip. Different marketing and promotion strategies can be formulated according to different hesitation degrees to promote the reduction of consumers' hesitation degree and influence the decision-making direction of consumers.

[0160] Consumer groups with search hesitation are more forward in the decision-making process. It can be considered to provide detailed brand and product comparison information in marketing and promotion activities, highlight the unique advantages of its own brand, provide more experience and trial services, and introduce price matching or other preferential policies.

[0161] B: As Figure 4 shown, the determination method of the hesitation degree of destination planning is as follows:

[0162] After the brand and category are selected, determine the hesitation degree of destination planning. Usually, the consumption of large commodities often involves planned visits to stores. Therefore, plans will be made in advance before traveling, planning what time and by what means to go to which stores.

[0163] b1: Analyze travel planning behavior based on point-of-sale dimension data, specifically including:

[0164] i. Multiple brands and multiple stores: Extremely hesitant;

[0165] Consumers are hesitant among a limited number of brand categories, but their willingness to travel to visit stores is already very strong.

[0166] ii. Single brand and multiple stores: Generally hesitant;

[0167] Consumers have already selected a brand and even had actual visits to multiple brands. They have initially determined the category, and the planning behavior may be comparing and screening distances, travel modes, and store visit sequences, and may even have entered the price comparison stage.

[0168] iii. Single brand and single store: Not hesitant;

[0169] Consumers' destinations are very clear, and the planning behavior may be to confirm the travel route or determine the public transportation transfer method, etc.

[0170] Analyze in combination with the planned travel mode:

[0171] Travel by car: Such consumers generally may already have a car, and the hesitation factor may be road conditions. This purchase may be a trade-in or an additional purchase;

[0172] Travel by bus: Such consumers may be making their first purchase, and the hesitation factors may be the number of transfers and the total travel time;

[0173] Walk / cycle: Such customers prefer stores at close range, but usually in the early stage of traveling to visit stores. Combining with subsequent planning data, if there is a plan for a store at a long distance of the same brand, it indicates that the decision-making stage is progressing continuously and is closer to the final decision;

[0174] b2: Judgment conclusion in the planning decision-making stage: The planning hesitation degree is divided into extremely hesitant, generally hesitant, and not hesitant. Sort according to the planning hesitation degree. The lower the hesitation degree, the closer to the final store selection decision. Different targeted marketing means can be used according to different hesitation degrees and consumers' target brands and stores to guide consumers' decisions.

[0175] C: As Figure 5 shown, the method for determining the hesitation degree of destination arrival is as follows:

[0176] The destination arrival hesitation degree is a comprehensive indicator that reflects the degree of uncertainty during the process of consumers from generating the purchase intention to actually visiting the store. This indicator is affected by multiple factors:

[0177] Planning cycle: The time length from the initial idea to the actual action;

[0178] Geographical factors: store location, transportation convenience, etc.

[0179] Destination search behavior: frequency and depth of online searches for store information;

[0180] Store route planning: select travel mode and plan route;

[0181] Actual travel behavior: whether the trip was made as planned, whether there were any additional visits, etc.

[0182] Usually when consumers decide to visit a store, their travel behavior consists of three actions: store search, store route planning, and travel navigation. First, determine the store's geographical location, then select a travel method (such as driving, public transportation, cycling, etc.) to plan the route, check the distance to the destination, the difficulty of reaching the destination, and determine the pre-selected travel route; finally, actually go to visit the store on the planned travel date.

[0183] Analyze the arrival hesitation of the destination based on the point of sale dimension data, including:

[0184] i. Arrive only at the planned store: no hesitation;

[0185] Characteristics: Consumers have clear goals and know what they want to buy; they may have done sufficient research;

[0186] Store visit behavior is highly consistent with expected plans;

[0187] Decision-making stage: The final decision is imminent and the purchase intention is strong;

[0188] Conversion rate expectation: High

[0189] ii. Not only arrived at the planned stores, but also visited other unplanned stores: generally hesitant;

[0190] Characteristics: Haven't made up their mind yet; want to make more comparisons in addition to the planned stores; decision may be affected by new information after field investigation;

[0191] Decision stage: close to the final decision, but still making final comparisons;

[0192] Conversion rate expectation: Medium

[0193] iii. Arrival at unplanned stores, not arriving at the originally planned stores, but arriving at other stores: particularly hesitant;

[0194] Characteristics: Consumers may only have preliminary interest; original plans may change due to various reasons (such as long distance, inconvenience in arrival); pre-selected brands may be changed;

[0195] Decision stage: Early in the decision stage, needs or preferences may change significantly;

[0196] Conversion rate expectation: low.

[0197] iv. No arrival, not reaching any store at all: extremely hesitant

[0198] Characteristics: may just be a spur-of-the-moment decision to view store information; have no strong intention to purchase, may just be a preliminary exploration; may be restricted by objective factors such as time and distance;

[0199] Decision-making stage: The earliest stage of the decision-making process, may not have entered the actual purchase consideration stage yet. Conversion rate expectation: extremely low.

[0200] c2: Judgment conclusion in the decision-making stage of destination arrival: The destination hesitation degree is divided into extremely hesitant, generally hesitant, and not hesitant. Sort according to the destination hesitation degree. The lower the hesitation degree, the closer to the final purchase decision. It can be based on different hesitation degrees and consumers.

[0201] Analysis from the time dimension: The longer the time interval from planning to arrival, usually the higher the hesitation degree; the in-store time on weekdays vs weekends may reflect different purchase urgencies, and the urgency on weekdays is higher;

[0202] Analysis from the spatial dimension: The closer the distance between the place of residence / work and the target store, the higher the convenience, and the hesitation degree may be lower; conversely, if still adhering to the original plan, it may indicate a strong purchase intention;

[0203] Analysis from the visit frequency: Multiple visits to the store may indicate strong interest and deeper hesitation;

[0204] Analysis from the number of store brands visited: Visiting the same brand may indicate a high degree of loyalty to the brand or having selected the brand, and it is very likely to be comparing prices and already very close to the final decision; Visiting stores of different brands indicates that the intended model may have been determined and the final comparison and decision are being made;

[0205] Marketing strategies: For the non-hesitant group: Strengthen the purchase confidence and provide a convenient transaction process; For the generally hesitant group: Provide detailed product comparison information and highlight unique advantages; For the extremely hesitant group: Increase brand exposure, provide more experience opportunities, and provide more selection and comparison tools;

[0206] D: As Figure 6 shown, the method for determining the hesitation degree of arriving at and then leaving the destination is as follows:

[0207] The hesitation degree model determines consumer behavior.

[0208] The destination hesitation degree not only includes behaviors in the store but also covers a series of behaviors after leaving the store, including visiting other stores and possible return visit behaviors.

[0209] Analyze the hesitation degree of consumers who arrive at and then leave the destination based on the point-of-sale dimension data, specifically including:

[0210] Analyze from the duration of consumers' stay in the store:

[0211] i. Stay in the store for a long time (more than 30 minutes), deeply understand the product, but finally do not purchase:

[0212] General hesitation

[0213] Characteristics: Show strong interest in the product; deeply understand the product; Decision-making stage: Close to purchase, but still have final concerns

[0214] Possibility of returning to the store: High

[0215] ii. Stay in the store for a general time (20 - 30 minutes), view the product, but leave without in-depth understanding: General hesitation

[0216] Characteristics: Have interest in the product, but not strong enough;

[0217] Decision-making stage: Still in the consideration stage, need more information or time

[0218] Possibility of returning to the store: Medium

[0219] iii. Stay in the store for a very short time (less than 20 minutes), quickly browse and then leave quickly: Special hesitation

[0220] Characteristics: Not very interested in the product; Leave quickly, seemingly losing interest;

[0221] Decision-making stage: May re-evaluate needs or consider other options

[0222] Possibility of returning to the store: Low

[0223] Analyze in combination with the behavior pattern of consumers after leaving the store:

[0224] iv. Return to the store again after leaving the store and there is no visit to other stores in between: Not hesitant;

[0225] The behavior of returning to the store again after leaving the store has positive significance. Especially in the case of no visit to other stores, it shows that the consumer is relatively satisfied with the visit experience of the target store, and the conversion probability is extremely high;

[0226] v. Visit 1 - 3 other stores after leaving the store: General hesitation;

[0227] vi. Visit more than 3 other stores after leaving the store: Special hesitation;

[0228] vii. Do not visit any store after leaving the store: Special hesitation;

[0229] Analyze in combination with the number of times consumers return to the store:

[0230] i. 1 - 2 times of returning to the store after leaving: without hesitation;

[0231] Consumers visit the store for the first time to learn about it.

[0232] According to the research results of automobile sales stores, the conversion rate of the first visit to the store is less than 10%, while the conversion rates of the 2nd and 3rd returns to the store are as high as 30%. As the number of returns to the store increases, the conversion rate decreases significantly, indicating that consumers maintain a relatively low level of hesitation during the first 3 visits. At this time, through appropriate marketing means, the probability of consumers making a purchase at the first destination is the highest;

[0233] ii. 3 times of returning to the store after leaving: generally hesitant;

[0234] iii. More than 3 times of returning to the store after leaving: extremely hesitant;

[0235] Conclusion on the judgment of the hesitation degree when leaving after arriving at the destination: The hesitation degree when leaving after arriving is divided into extremely hesitant, generally hesitant, and without hesitation. Sorted according to the destination hesitation degree, the lower the hesitation degree, the higher the conversion rate.

[0236] Suggestions for marketing strategies:

[0237] Group without hesitation: Provide discounts or time - limited promotions to facilitate quick decision - making; Strengthen product advantages and provide exclusive customer services; Simplify the purchase process and provide convenient payment options;

[0238] Group with general hesitation: Provide detailed product comparison information; Arrange in - depth experience activities; Provide flexible purchase plans;

[0239] Group with extreme hesitation: Increase brand exposure and provide basic education content; Provide stress - free consultation services;

[0240] Invite to participate in brand activities to enhance emotional connection.

[0241] In other embodiments of the present invention, there is also provided a device for running the above - mentioned method, such as Figure 7 shown, including:

[0242] A data acquisition module, used to acquire data, and the data includes consumers' online activity data and offline navigation software usage data, and determine the selling point spatial location and product brand information from the acquired data;

[0243] A data pre - processing module, used to pre - process the acquired data to form selling point dimension data;

[0244] A data mining module, used to mine the correlation and mutual influence between the selling point dimension data, and predict the consumption probability based on hesitation degree analysis.

[0245] In other embodiments of the present invention, a computer-readable storage medium for running the above method is also provided, etc.

[0246] The units involved in the embodiments of the present invention can be implemented in software or in hardware. The described units can also be provided in a processor. Among them, the names of these units do not constitute a limitation to the unit itself in some cases.

[0247] According to one aspect of the embodiments of the present invention, a computer program product or a computer program is provided. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the methods provided in the above various alternative implementation manners.

[0248] As another aspect, the embodiments of the present invention also provide a computer-readable medium. The computer-readable medium can be included in the electronic device described in the above embodiments; or it can exist separately and is not assembled into the electronic device. The above computer-readable medium carries one or more programs. When the one or more programs are executed by an electronic device, the electronic device implements the methods described in the above embodiments.

Claims

1. A consumption probability prediction method based on hesitation degree analysis, characterized in that The following steps are involved: Collect data, including consumer online activity data and offline navigation software usage data, and determine the point-of-sale space location and product brand information from the collected data; The collected data is preprocessed based on a computer processor running a program to form point-of-sale dimension data; the preprocessing of the collected data to form the point-of-sale dimension data specifically includes sub-steps: preprocessing the collected data by cleaning, deduplication, and time alignment to ensure the consistency and comparability of the data, and finally associating the destination data with the brand, category, and surrounding environment to form the data as the point-of-sale dimension data; The preprocessing of the collected data specifically includes preprocessing of basic information data of the point of sale, and the preprocessing of basic information data of the point of sale specifically includes sub-steps: a) Use the spatial search interface to search for keywords in the city where the store is located using the permutations and combinations of brand, name, and address. Each spatial query will return multiple results. The IDs obtained from all permutations and combinations are summarized and counted for their occurrences. b) Sort the aggregated IDs by the number of occurrences, select the number of occurrences of the first set value of the ID in the ranking for comparison, if the number of occurrences of the first place is ≥ the second set value, it is considered that the ID of this store in the spatial data of the professional map provider is this ID, otherwise, the store address is reverse geocoded and the distance is compared with the coordinates corresponding to the ID, and a comprehensive score is given based on the distance between the reverse geocoding and the coordinates corresponding to the ID and the number of ID occurrences, and finally the one with the highest score is selected as the store ID; c) Check the ID determined by distance comparison on the maps of each map provider and make corrections based on the check results; d) Compare the basic information data of the sales points collected by the vertical media platform to determine their relevance, and then complete the missing information on the official website, including the categories and models of products sold; Based on a computer processor running a program, the correlation and mutual influence between the sales point dimension data are mined, and a consumer hesitation analysis model is constructed, which specifically includes destination search hesitation, destination planning hesitation, destination arrival hesitation and destination arrival and departure hesitation, and consumption probability prediction is performed based on the hesitation analysis model.

2. The consumption probability prediction method based on hesitation degree analysis according to claim 1, wherein The method of determining the spatial location of the point of sale from the collected data specifically includes the following sub-steps: searching the collected data through the Internet, cross-checking and comparing the names disclosed on the official website and the recommended names to determine the accurate name of the point of sale and the longitude and latitude coordinates, and finally circling the target location, adjusting the range, and cross-checking with third-party data to finally determine the spatial location of the point of sale.

3. The consumption probability prediction method based on hesitation degree analysis according to claim 2, wherein The consumer's online activity data includes search records, browsing history and consultation content; the offline navigation software usage data includes departure location, arrival time, route selection, entry and exit time, length of stay, portrait information, arrival method, corresponding sales point search, viewing, route planning and navigation travel data; the product brand information includes independent sub-brands; the adjustment range specifically includes the capture range of visit behavior of the designated sales point destination, which includes the store showroom and public areas outside the store.

4. The consumption probability prediction method based on hesitation degree analysis according to claim 1, wherein The preprocessing of the collected data specifically includes arrival data preprocessing, and in the arrival data preprocessing, it specifically includes sub-steps: The arrival data includes the stay duration and the entry and exit times; within the capture range, the visit records with a stay duration less than the set value and entry and exit times outside the store business hours are all excluded as abnormal data. If the stay duration is a null value, the mode is used for filling; Dimensionality reduction according to the arrival time: According to the passenger flow curve of the commodity store, the arrival data is dimensionally reduced into entering the store during the peak period and entering the store during the non-peak period; Dimensionality reduction according to the arrival date: According to the characteristics of arriving at the store on weekdays and holidays, the arrival data is divided into arriving at the store on weekdays and non-weekdays; Dimensionality reduction according to the accuracy of arrival judgment: According to the way of arriving at the store, it is distinguished between accurate arrival at the store and inaccurate arrival at the store; The distinguishing method includes: If the navigation destination is consistent with the store entering the capture fence, it is regarded as accurate arrival at the store, otherwise it is regarded as inaccurate arrival at the store.

5. The consumption probability prediction method based on hesitation degree analysis according to claim 1, characterized in that The comparison with the basic information data of the sales points collected by the vertical media platform is used to determine its relevance, and then the information missing from the official website is supplemented, which specifically includes sub-steps: 1), Normalize and standardize the name and address; 2), Use the similarity calculation and comparison of the address and name under the same province and city, and distinguish the situations of absolute coincidence, very coincidence, need for spot check and need for manual intervention judgment according to the score.

6. The consumption probability prediction method based on hesitation degree analysis according to claim 1, wherein The method for determining the hesitation degree of destination retrieval specifically includes the following sub-steps: a1: Obtain the user's true destination from the preprocessed data; a2: Disassociate and compare the destination with the category and brand; a3: Analyze the retrieval behavior from the sales point dimension data; a4: Based on the analysis of the retrieval behavior, form a judgment conclusion at the decision-making stage of the retrieval: Divide the hesitation degree of the retrieval into especially hesitant, generally hesitant, and not hesitant, and sort according to the retrieval hesitation degree. The lower the hesitation degree, the closer to the final brand category and the decision of the planned trip.

7. The consumption probability prediction method based on hesitation degree analysis according to claim 1, characterized in that The method for determining the hesitation degree of destination planning specifically includes the following sub-steps: b1: Analyze the travel planning behavior according to the sales point dimension data; b2: Based on the analysis of the planning behavior, form a judgment conclusion at the decision-making stage of the planning: The hesitation degree of destination planning is divided into especially hesitant, generally hesitant, and not hesitant, and sort according to the planning hesitation degree. The lower the hesitation degree, the closer to the final store selection decision.

8. The consumption probability prediction method based on hesitation degree analysis according to claim 1, characterized in that The method for determining the hesitation degree of destination arrival specifically includes the following sub-steps: c1: Analyze the hesitation degree of destination arrival from the sales point dimension data; c2: Based on the analysis of the hesitation degree of whether the destination is reached, form a judgment conclusion at the decision-making stage of destination arrival: The hesitation degree of destination arrival is divided into especially hesitant, generally hesitant, and not hesitant, and sort according to the destination hesitation degree. The lower the hesitation degree, the closer to the final purchase decision.

9. The consumption probability prediction method based on hesitation degree analysis according to claim 1, wherein The method for determining the hesitation degree of arriving at the destination and then leaving specifically includes the following sub-steps: d1: Analyze the hesitation degree of arriving at the destination and then leaving from the sales point dimension data; d2: Based on the analysis of the consumer's stay duration in the store, form a judgment conclusion: Divide the hesitation degree of leaving after arrival into especially hesitant, generally hesitant, and not hesitant, and sort according to the destination hesitation degree. The lower the hesitation degree, the higher the conversion rate.

10. The consumption probability prediction method based on hesitation degree analysis according to claim 1, characterized in that In the method for determining the hesitation degree of the destination arriving and then leaving, the following sub-steps are further included: Combining the analysis of the post-store behavior patterns of consumers, specifically including the following situations: If the consumer returns to the store after leaving the store and there is no visit to other stores in between, it is judged as not hesitant; If the consumer visits 1 - 3 stores after leaving the store, it is judged as generally hesitant; If the consumer visits more than 3 stores or does not visit any store after leaving the store, it is judged as particularly hesitant.

11. The consumption probability prediction method based on hesitation degree analysis according to claim 1, characterized in that In the method for determining the hesitation degree of the destination arriving and then leaving, the following sub-steps are further included: Combining the analysis of the number of times the consumer returns to the store, specifically including the following situations: Returning to the store 1 - 2 times after leaving the store, it is judged as not hesitant; Returning to the store 3 times after leaving the store, it is judged as generally hesitant; Returning to the store more than 3 times after leaving the store: particularly hesitant.

12. The consumption probability prediction method based on hesitation degree analysis according to claim 1, characterized in that After predicting the consumption probability by using the analysis of the consumer's hesitation degree, the following steps are further included: Formulating different marketing and promotion strategies according to different hesitation levels to promote the reduction of the consumer's hesitation degree and influence the consumer's decision-making direction, specifically including the following situations: Providing discounts or time-limited promotions to the non-hesitant group to promote quick decision-making, strengthening the product advantages, providing exclusive customer services, and simplifying the purchase process, providing convenient payment options; Providing detailed product comparison information, arranging in-depth experience activities, and providing flexible purchase plans to the generally hesitant group; For the particularly hesitant group, increasing brand exposure, providing basic education content, providing stress-free consultation services, inviting them to participate in brand activities, and enhancing emotional connection.

13. A consumption probability prediction device based on hesitation degree analysis, characterized in that, Including: A data collection module, used to collect data, and the data includes consumers' online activity data and offline navigation software usage data, and determining the sales point spatial location and commodity brand information from the collected data; A data preprocessing module, used to preprocess the collected data based on a program running on a computer processor to form sales point dimension data; The preprocessing of the collected data to form sales point dimension data specifically includes sub-steps: Performing preprocessing operations such as cleaning, deduplication, and time alignment on the collected data to ensure the consistency and comparability of the data, and finally using the data obtained by associating the destination data with the brand, category, and surrounding environment as the sales point dimension data; The preprocessing of the collected data specifically includes the preprocessing of the basic information data of the sales point. In the preprocessing of the basic information data of the sales point, the following sub-steps are specifically included: a), Using the permutation and combination of the brand, name, and address to perform keyword searches within the scope of the store's city using the spatial search interface. Each spatial query of the combination will obtain multiple return results, and the IDs obtained from all permutations and combinations are summarized and the number of times they appear is statistically counted; b), Sorting the summarized IDs according to the number of times they appear, selecting the number of times the ID with the top first set value appears for comparison. If the number of times the first one appears ≥ the second set value, it is considered that the ID of this store in the spatial data of the professional mapping company is this ID. Otherwise, the address of the store is reverse geocoded, and the distance is compared with the coordinates corresponding to the ID. A comprehensive score is calculated based on the distance between the reverse geocoding and the coordinates corresponding to the ID and the number of times the ID appears, and finally the one with the highest score is selected as the ID of the store; c), check the ID determined by distance comparison on the maps of each map merchant and make corrections according to the check results; d), compare with the basic information data of the sales points collected by the vertical media platform to determine its relevance, and then supplement the information missing on the official website, including the sold product categories and models; A data mining module, which is used to mine the relevance and mutual influence between the sales point dimension data based on the operation of a computer processor program, and construct a consumer hesitation analysis model, specifically including destination retrieval hesitation, destination planning hesitation, destination arrival hesitation, and hesitation of leaving after arriving at the destination, and predict the consumption probability based on the hesitation analysis model.

14. A computer-readable storage medium, characterized in that, A computer program is stored in a readable storage medium, and when the computer program is loaded by a processor, it executes the method according to any one of claims 1 to 5.

15. A consumption probability prediction system based on hesitation degree analysis, characterized in that, It includes the consumption probability prediction device based on hesitation analysis according to claim 13.

Images