Charging Customer Conversion Propensity Prediction Methods and Related Products

Abstract

This invention discloses a method for predicting customer conversion propensity for recharging services and related products. The method predicts the recharging consumption behavior of non-converted customers, then selects potential conversion customers, performs three clustering operations, and finally obtains the predicted purchase probability of potential conversion customers for each type of non-recharging consumption. This method enables a complete process from source data to predictive analysis to formalized output, with results directly meeting multi-faceted business needs and allowing business personnel to use it directly without coding.

Description

Technical Field

[0001] This invention relates to a method for predicting customer conversion propensity for charging services and related products. Background Technology

[0002] Customers refueling at gas stations may purchase other goods afterward. Similarly, customers refueling with natural gas may purchase other non-oil / gas products. Accurate prediction of customers' non-oil / gas consumption tendencies is necessary to provide reliable supplementary information for marketers. Summary of the Invention

[0003] This invention provides a method and related products for predicting the conversion tendency of charging customers, in order to obtain rich consumption tendency information of charging customers.

[0004] The technical solution of the present invention is as follows: A method for predicting the conversion tendency of charging customers, comprising:

[0005] Input the charging consumption characteristic time series of unconverted customers within a first set time period up to the current period into a pre-trained prediction model to obtain the predicted charging consumption characteristic time series of unconverted customers for a second set time period after the current period. The prediction model is trained by using the charging consumption characteristic time series of converted customers for the second set time period before their first non-charging consumption as the output and the charging consumption characteristic time series of the first set time period before that as the input.

[0006] The charging consumption characteristic time series of unconverted customers within the first set time period up to the current time period is concatenated with the predicted charging consumption characteristic time series of unconverted customers after the current time period in the second set time period to obtain the concatenated charging consumption characteristic time series. The similarity between the concatenated charging consumption characteristic time series and the charging consumption characteristic time series of converted customers before their first non-charging consumption in the first set time period and the sum of the first and second set time periods is calculated. Unconverted customers with a similarity greater than a set threshold are selected as potential converted customers, and the corresponding similarity is used as the shopping probability of potential converted customers.

[0007] Cluster the charging consumption characteristics time series of potential converted customers within the first set time period up to the current period and the charging consumption characteristics time series of converted customers before the first non-charging consumption within the second set time period before the first set time period, to obtain the first cluster center and the set number of sample points closest to the first cluster center;

[0008] Cluster the predicted charging consumption characteristics time series of potential converted customers for a second set period after the current period and the charging consumption characteristics time series of converted customers for a second set period before their first non-charging consumption, to obtain the second cluster center and the set number of sample points closest to the second cluster center;

[0009] Cluster the spliced ​​charging consumption characteristic time series of potential converted customers and the charging consumption characteristic time series of converted customers before their first non-charging consumption, which is the sum of the first and second set durations, to obtain the center of the third cluster and the set number of sample points closest to the third cluster.

[0010] Based on the frequency of various non-charging consumption by converted customers and the distance between potential converted customers and the first, second, and third cluster centers, predict the predicted purchase probability of various non-charging consumption by potential converted customers.

[0011] Optionally, it also includes: normalizing the predicted purchase probability of various non-charging consumptions of potential converted customers, and outputting the maximum normalized value of potential converted customers and the corresponding non-charging consumption type.

[0012] Alternatively, the normalization formula is as follows: l i =p i (Sigmoid((Max(r ij )-r kj ) / (Max(r ij )-Min(r ij )))),l i p is the maximum normalized value of potential converting customer i. i Let r be the purchase probability of potential customer i, sigmoid be the activation function, Max be the maximum value function, and Min be the minimum value function. ij It represents the predicted purchase probability of potential customer i in non-charging consumption category j.

[0013] Optionally, it also includes: outputting the shopping probability of potential converting customers.

[0014] Optionally, the predicted purchase probability of various types of non-energized consumption by potential converted customers is equal to the frequency of the corresponding type of non-energized consumption of converted customers multiplied by the sum of the distances between potential converted customers and the first, second, and third cluster centers.

[0015] Optionally, the prediction model is a bidirectional long short-term memory neural network.

[0016] Optionally, the charging customer is a refueling customer or a gas refueling customer.

[0017] The technical solution of the present invention is as follows: A charging customer conversion tendency prediction device, comprising:

[0018] The prediction module is used to input the charging consumption characteristic time series of unconverted customers within a first set time period up to the current time period into a pre-trained prediction model to obtain the predicted charging consumption characteristic time series of unconverted customers for a second set time period after the current time period. The prediction model is trained by using the charging consumption characteristic time series of converted customers for the second set time period before their first non-charging consumption as the output and the charging consumption characteristic time series of the first set time period before that as the input.

[0019] The filtering module is used to concatenate the charging consumption feature time series of unconverted customers within a first set time period up to the current time period with the predicted charging consumption feature time series of unconverted customers after the current time period for a second set time period, to obtain a concatenated charging consumption feature time series. The similarity between the concatenated charging consumption feature time series and the charging consumption feature time series of converted customers before their first non-charging consumption for the sum of the first and second set time periods is calculated. Unconverted customers with a similarity greater than a set threshold are selected as potential converted customers, and the corresponding similarity is used as the shopping probability of potential converted customers.

[0020] The first clustering module is used to cluster the charging consumption characteristic time series of potential converted customers within a first set time period up to the current period and the charging consumption characteristic time series of converted customers before the first non-charging consumption within a second set time period before the first set time period, to obtain the first cluster center and the set number of sample points closest to the first cluster center.

[0021] The second clustering module is used to cluster the predicted charging consumption characteristic time series of potential converted customers for a second set period after the current period and the charging consumption characteristic time series of converted customers for a second set period before their first non-charging consumption, to obtain the second cluster center and the set number of sample points closest to the second cluster center.

[0022] The third clustering module is used to cluster the spliced ​​charging consumption feature time series of potential converted customers and the charging consumption feature time series of converted customers before their first non-charging consumption, which is the sum of the first set time and the second set time. The result is to obtain the center of the third cluster and the set number of sample points closest to the third cluster.

[0023] The prediction module is used to predict the purchase probability of various non-charging consumption of potential customers based on the frequency of various non-charging consumption of converted customers and the distance between potential converted customers and the first, second and third cluster centers.

[0024] The technical solution of the present invention is as follows: a charging customer conversion tendency prediction device, comprising: a memory and a processor, wherein a program is stored in the memory, and the processor runs the program to execute the above-mentioned charging customer conversion tendency prediction method.

[0025] The technical solution of the present invention is as follows: a program product that executes the above-mentioned charging customer conversion tendency prediction method when running on a processor.

[0026] By simply providing the time series of charging consumption characteristics of converted and non-converted customers, as well as the time series of non-charging consumption characteristics of converted customers, it is possible to predict non-converted customers and determine the predicted purchase probability of different types of non-charging products for potential converted customers. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating the charging customer conversion tendency prediction method of the present invention.

[0028] Figure 2 This is a structural block diagram of the charging customer conversion tendency prediction device of the present invention.

[0029] Figure 3 This is another structural block diagram of the charging customer conversion tendency prediction device of the present invention. Detailed Implementation

[0030] The present invention will be further described below with reference to specific embodiments, but the scope of protection of the present invention is not limited thereto.

[0031] refer to Figure 1 The present invention provides a method for predicting the conversion tendency of charging customers, including the following steps.

[0032] Step 101: Input the charging consumption characteristic time series of the unconverted customers within the first set time period up to the current time period into the pre-trained prediction model to obtain the predicted charging consumption characteristic time series of the unconverted customers for the second set time period after the current time period. The prediction model is trained by using the charging consumption characteristic time series of the converted customers for the second set time period before their first non-charging consumption as the output and the charging consumption characteristic time series of the first set time period before that as the input.

[0033] The following explanation uses refueling consumption at gas stations as an example. The table below provides an exemplary representation of a single data point in a time series of a specific customer's refueling consumption characteristics.

[0034] uid Unified User Identity 9sadjfb723r timestamp Timestamp 20210601 age age 31 sex gender male type Oil type 95 litter Refueling and increasing numbers 5 amount fuel amount 300 station Gas station name A gas station

[0035] The table below is an example of a data point in a time series of non-charging consumption characteristics of a specific converted customer.

[0036]

[0037]

[0038] Each data point above corresponds to a time slice, for example, 1 day. The first set duration is, for example, 100 days, and the second set duration is, for example, 10 days. Based on the fuel consumption data of unconverted customers in the most recent 100 days, predict the fuel consumption data of unconverted customers in the next 10 days.

[0039] The time slice can be measured in terms of a week or a month.

[0040] The predictive model is trained using refueling data from converted customers, particularly by extracting correlations between refueling behavior characteristics of converted customers before their first non-refueling purchase and their previous refueling behavior characteristics. The entire process of the method is driven by known historical data.

[0041] Step 102: Concatenate the charging consumption feature time series of unconverted customers within the first set duration up to the current time period with the predicted charging consumption feature time series of unconverted customers after the current time period for the second set duration. This results in a concatenated charging consumption feature time series. Calculate the similarity between the concatenated charging consumption feature time series and the charging consumption feature time series of converted customers before their first non-charging consumption, which is the sum of the first and second set durations. Select unconverted customers with a similarity greater than a set threshold as potential converted customers, and use the corresponding similarity as the shopping probability of the potential converted customers.

[0042] Continuing with the previous example, we compare the refueling data of unconverted customers over the past 100 days and the predicted data for the next 10 days with the refueling data of converted customers over the 110 days prior to their first non-refueling purchase. Similarity can be measured using Euclidean distance. A threshold can be set, such as 0.6 or 0.7, or any other arbitrary threshold. A higher similarity score indicates that the refueling consumption behavior of unconverted and converted customers is more similar.

[0043] Step 103: Cluster the charging consumption characteristic time series of potential converted customers within the first set time period up to the current period and the charging consumption characteristic time series of converted customers before the first non-charging consumption within the second set time period before the first set time period, to obtain the first cluster center and the set number of sample points closest to the first cluster center.

[0044] Any known clustering algorithm can be used. Examples of clustering algorithms include K-means clustering and K-means++ clustering.

[0045] Following the previous example, we clustered the refueling data of potential customers over the past 100 days with the refueling data of converted customers from 110 days before their first non-refueling purchase to 10 days before their first non-refueling purchase.

[0046] Step 104: Cluster the predicted charging consumption characteristic time series of potential converted customers for a second set duration after the current time period and the charging consumption characteristic time series of converted customers for a second set duration before their first non-charging consumption, to obtain the second cluster center and the set number of sample points closest to the second cluster center.

[0047] Following the previous example, we clustered the fuel data of potential customers predicted for the next 10 days with the fuel data of converted customers in the 10 days prior to their first non-fuel purchase.

[0048] Step 105: Cluster the spliced ​​charging consumption feature time series of potential converted customers and the charging consumption feature time series of converted customers before their first non-charging consumption, which is the sum of the first set duration and the second set duration, to obtain the center of the third cluster and the set number of sample points closest to the third cluster.

[0049] Following the previous example, we clustered the refueling data of potential customers over the past 100 days and the next 10 days with the refueling data of converted customers over the 110 days prior to their first non-refueling purchase.

[0050] The above analysis examines the refueling behavior of potential and converted customers from three different dimensions, analyzing the degree of similarity between the two.

[0051] Step 106: Based on the frequency of various non-charging consumption of converted customers and the distance between potential converted customers and the first, second, and third cluster centers, predict the predicted purchase probability of various non-charging consumption of potential converted customers.

[0052] The product categories are categorized based on the frequency of food consumption, medicine consumption, and toilet paper consumption of converted customers, and statistics are compiled based on the consumption records of converted customers. For example, the frequency of food consumption for converted customers is once a week, and the frequency of medicine consumption is 0.1 times a week.

[0053] For example, the predicted probability of a certain potential customer purchasing food within the next 10 days is 0.7, and the predicted probability of purchasing medicine within the next 10 days is 0.1.

[0054] Optionally, the predicted purchase probability of various types of non-energized consumption by potential converted customers is equal to the frequency of the corresponding type of non-energized consumption of converted customers multiplied by the sum of the distances between potential converted customers and the first, second, and third cluster centers.

[0055] When a potential customer is not located near any first-class cluster center, the distance between its data point and the cluster center can be set to a larger value.

[0056] The distance between a potential converting customer's data point and the cluster center can be measured using Euclidean distance.

[0057] Optionally, the method further includes: normalizing the predicted purchase probabilities of various non-charging consumption types for potential converted customers, and outputting the maximum normalized value of the potential converted customer and the corresponding non-charging consumption type. Normalization is performed to ensure that the displayed data has a uniform standard.

[0058] Optionally, the normalization formula is as follows: li=pi(Sigmoid((Max(rij)-rkj) / (Max(rij)-Min(rij)))), where li is the maximum normalized value of potential customer i, pi is the purchase probability of potential customer i, sigmoid is the activation function, Max is the maximum value function, Min is the minimum value function, and rij is the predicted purchase probability of potential customer i in non-energy consumption of type j.

[0059] Optionally, it also includes: outputting the shopping probability of potential converting customers.

[0060] In one embodiment, the method output is as follows: the probability of potential customer a making a non-fuel purchase within the next 10 days is 0.5 (i.e., the pi value), and the predicted purchase probability of him making a food purchase (i.e., number j) is the highest, with a predicted purchase probability of 0.5 for food purchase.

[0061] Optionally, the prediction model is a bidirectional long short-term memory neural network. The prediction model needs to uncover the causal relationships between data sequences and can employ any known and appropriate model.

[0062] Optionally, the charging customer is a refueling customer or a gas refueling customer.

[0063] refer to Figure 2 Based on the same inventive concept, embodiments of the present invention also provide a charging customer conversion tendency prediction device, comprising:

[0064] Prediction module 1 is used to input the charging consumption characteristic time series of unconverted customers within a first set time period up to the current time period into a pre-trained prediction model to obtain the predicted charging consumption characteristic time series of unconverted customers for a second set time period after the current time period. The prediction model is trained by using the charging consumption characteristic time series of converted customers for the second set time period before their first non-charging consumption as the output and the charging consumption characteristic time series of the first set time period before that as the input.

[0065] The filtering module 2 is used to concatenate the charging consumption feature time series of unconverted customers within a first set time period up to the current time period with the predicted charging consumption feature time series of unconverted customers after the current time period for a second set time period, to obtain a concatenated charging consumption feature time series. The similarity between the concatenated charging consumption feature time series and the charging consumption feature time series of converted customers before their first non-charging consumption for the sum of the first set time period and the second set time period is calculated. Unconverted customers with a similarity greater than a set threshold are selected as potential converted customers, and the corresponding similarity is used as the shopping probability of potential converted customers.

[0066] The first clustering module 3 is used to cluster the charging consumption characteristic time series of potential converted customers within a first set time period up to the current period and the charging consumption characteristic time series of converted customers before the first non-charging consumption within a second set time period before the first set time period, to obtain the first cluster center and the set number of sample points closest to the first cluster center.

[0067] The second clustering module 4 is used to cluster the predicted charging consumption characteristic time series of potential converted customers for a second set time period after the current time period and the charging consumption characteristic time series of converted customers for a second set time period before the first non-charging consumption, to obtain the second cluster center and the set number of sample points closest to the second cluster center.

[0068] The third clustering module 5 is used to cluster the spliced ​​charging consumption feature time series of potential converted customers and the charging consumption feature time series of converted customers before their first non-charging consumption, which is the sum of the first set time and the second set time, to obtain the center of the third cluster and the set number of sample points closest to the third cluster.

[0069] Prediction module 6 is used to predict the purchase probability of various non-charging consumption of potential customers based on the frequency of various non-charging consumption of converted customers and the distance between potential converted customers and the first, second and third cluster centers.

[0070] Optionally, the prediction module 6 is also used to normalize the predicted purchase probability of various non-charging consumption of potential converted customers, and output the maximum normalized value of potential converted customers and the corresponding non-charging consumption type.

[0071] Optionally, prediction module 6 is also used to output the shopping probability of potential converting customers.

[0072] refer to Figure 3 Based on the same inventive concept, embodiments of the present invention also provide a charging customer conversion tendency prediction device, comprising: a memory and a processor, wherein a program is stored in the memory and the processor runs the program to perform the above-described charging customer conversion tendency prediction method.

[0073] The memory can be any known type of memory, such as a hard drive, flash memory, or optical disc. The processor can be any known type of processor, such as a central processing unit (CPU) or a graphics processing unit (GPU).

[0074] Based on the same inventive concept, embodiments of the present invention also provide a program product that executes the above-described charging customer conversion tendency prediction method when running on a processor.

[0075] The various embodiments in this invention are described in a progressive manner. For the same or similar parts between the various embodiments, please refer to each other. Each embodiment focuses on describing the differences from other embodiments.

[0076] The scope of protection of this invention is not limited to the embodiments described above. Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its scope and spirit. If these modifications and variations fall within the scope of the claims of this invention and their equivalents, then the intent of this invention also includes these modifications and variations.

Claims

1. A method for predicting customer conversion propensity for charging services, characterized in that, include: Input the charging consumption characteristic time series of unconverted customers within a first set time period up to the current period into a pre-trained prediction model to obtain the predicted charging consumption characteristic time series of unconverted customers for a second set time period after the current period. The prediction model is trained by using the charging consumption characteristic time series of converted customers for the second set time period before their first non-charging consumption as the output and the charging consumption characteristic time series of the first set time period before that as the input. The charging consumption characteristic time series of unconverted customers within the first set time period up to the current time period is concatenated with the predicted charging consumption characteristic time series of unconverted customers after the current time period in the second set time period to obtain the concatenated charging consumption characteristic time series. The similarity between the concatenated charging consumption characteristic time series and the charging consumption characteristic time series of converted customers before their first non-charging consumption in the first set time period and the sum of the first and second set time periods is calculated. Unconverted customers with a similarity greater than a set threshold are selected as potential converted customers, and the corresponding similarity is used as the shopping probability of potential converted customers. Cluster the charging consumption characteristics time series of potential converted customers within the first set time period up to the current period and the charging consumption characteristics time series of converted customers before the first non-charging consumption within the second set time period before the first set time period, to obtain the first cluster center and the set number of sample points closest to the first cluster center; Cluster the predicted charging consumption characteristics time series of potential converted customers for a second set period after the current period and the charging consumption characteristics time series of converted customers for a second set period before their first non-charging consumption, to obtain the second cluster center and the set number of sample points closest to the second cluster center; Cluster the spliced ​​charging consumption characteristic time series of potential converted customers and the charging consumption characteristic time series of converted customers before their first non-charging consumption, which is the sum of the first and second set durations, to obtain the center of the third cluster and the set number of sample points closest to the third cluster. Based on the frequency of various non-charging consumption of converted customers and the distance between potential converted customers and the first, second and third cluster centers, predict the predicted purchase probability of various non-charging consumption of potential converted customers. Among them, the predicted purchase probability of various non-energized consumption of potential converted customers is equal to the frequency of corresponding non-energized consumption of converted customers multiplied by the sum of the distances between potential converted customers and the first, second, and third cluster centers.

2. The method according to claim 1, characterized in that, Also includes: The predicted purchase probability of various non-charging consumptions of potential customers is normalized, and the maximum normalized value of potential customers and the corresponding non-charging consumption type are output.

3. The method according to claim 2, characterized in that, The formula for normalization is as follows: l i =p i (Sigmoid((Max(r ij )-r ij ) / (Max(r ij )-Min(r ij )))),l i p is the maximum normalized value of potential converting customer i. i Let r be the purchase probability of potential customer i, sigmoid be the activation function, Max be the maximum value function, and Min be the minimum value function. ij It represents the predicted purchase probability of potential customer i in non-charging consumption category j.

4. The method according to claim 1, characterized in that, Also includes: Output the probability of a potential customer making a purchase.

5. The method according to claim 1, characterized in that, The prediction model is a bidirectional long short-term memory neural network.

6. The method according to claim 1, characterized in that, The customers who charge the gas are either refueling customers or gas refueling customers.

7. A device for predicting customer conversion propensity for charging services, characterized in that, include: The prediction module is used to input the charging consumption characteristic time series of unconverted customers within a first set time period up to the current time period into a pre-trained prediction model to obtain the predicted charging consumption characteristic time series of unconverted customers for a second set time period after the current time period. The prediction model is trained by using the charging consumption characteristic time series of converted customers for the second set time period before their first non-charging consumption as the output and the charging consumption characteristic time series of the first set time period before that as the input. The filtering module is used to concatenate the charging consumption feature time series of unconverted customers within a first set time period up to the current time period with the predicted charging consumption feature time series of unconverted customers after the current time period for a second set time period, to obtain a concatenated charging consumption feature time series. The similarity between the concatenated charging consumption feature time series and the charging consumption feature time series of converted customers before their first non-charging consumption for the sum of the first and second set time periods is calculated. Unconverted customers with a similarity greater than a set threshold are selected as potential converted customers, and the corresponding similarity is used as the shopping probability of potential converted customers. The first clustering module is used to cluster the charging consumption characteristic time series of potential converted customers within a first set time period up to the current period and the charging consumption characteristic time series of converted customers before the first non-charging consumption within a second set time period before the first set time period, to obtain the first cluster center and the set number of sample points closest to the first cluster center. The second clustering module is used to cluster the predicted charging consumption characteristic time series of potential converted customers for a second set period after the current period and the charging consumption characteristic time series of converted customers for a second set period before their first non-charging consumption, to obtain the second cluster center and the set number of sample points closest to the second cluster center. The third clustering module is used to cluster the spliced ​​charging consumption feature time series of potential converted customers and the charging consumption feature time series of converted customers before their first non-charging consumption, which is the sum of the first set time and the second set time. The result is to obtain the center of the third cluster and the set number of sample points closest to the third cluster. The prediction module is used to predict the purchase probability of various non-charging consumption of potential customers based on the frequency of various non-charging consumption of converted customers and the distance between potential converted customers and the first, second and third cluster centers. Among them, the predicted purchase probability of various non-energized consumption of potential converted customers is equal to the frequency of corresponding non-energized consumption of converted customers multiplied by the sum of the distances between potential converted customers and the first, second, and third cluster centers.

8. A device for predicting customer conversion propensity for charging services, characterized in that, include: A memory and a processor, wherein a program is stored in the memory and the processor runs the program to perform the charging customer conversion tendency prediction method according to any one of claims 1 to 6.

9. A program product, characterized in that, When it runs on a processor, it executes the charging customer conversion tendency prediction method according to any one of claims 1 to 6.

Images