A fixed and mobile express cabinet user demand prediction method

Abstract

The application discloses a kind of fixed and mobile express cabinet user demand prediction method, first, the gender, age, employment status, education level, income and monthly online shopping frequency of the user in target area are collected Social demographic attribute data and data cleaning is carried out;Prefer to analyze, determine the experimental method and carry out attribute division and classification;Finally, the model is established to estimate parameters and apply.The method of the application reveals the influencing factors affecting the user's choice of mobile express cabinet or fixed express cabinet, which is beneficial to reveal the true potential of introducing mobile express cabinet service and help policymakers better grasp the positioning of mobile express cabinet in the initial stage.

Description

Technical Field

[0001] This invention relates to the field of express parcel storage, and more specifically to a method for predicting user demand for fixed and mobile express lockers. Background Technology

[0002] Due to the rapid development of the e-commerce market, a large number of parcels are delivered in urban areas every day, and door-to-door delivery is considered the most common last-mile delivery method. However, traditional door-to-door delivery is inefficient, which means that last-mile delivery accounts for more than 40% of the total cost of the entire delivery process, leading to problems such as congestion, air pollution, and greenhouse gas emissions.

[0003] To address the challenges posed by last-mile delivery, several innovative and cost-effective last-mile models have emerged, one of which is the use of parcel lockers. Parcel lockers can be categorized into two types: fixed parcel lockers (SPL) and mobile parcel lockers (MPL). The potential benefits and advantages of parcel locker models largely depend on customer preferences related to the two different models.

[0004] Therefore, this invention aims to explore customer preference in the context of the coexistence of two parcel locker modes, and proposes a method for predicting user demand for fixed and mobile parcel lockers. Summary of the Invention

[0005] The purpose of this invention is to provide a method for predicting user demand for fixed and mobile parcel lockers. Based on the socio-demographic attributes of users in the target area, the service attributes and contextual attributes of fixed and mobile parcel lockers, the method reveals the potential for introducing mobile parcel locker services, which helps policymakers better grasp the positioning of the mobile parcel locker model during the project initiation phase.

[0006] To achieve the above functions, this invention designs a method for predicting the demand of users for fixed and mobile parcel lockers. For users in a target area, the following steps S1-S3 are executed to complete the prediction of user demand for fixed and mobile parcel lockers:

[0007] Step S1: Collect socio-demographic attribute data of users in the target area and preprocess the socio-demographic attribute data of users;

[0008] Step S2: For each attribute of fixed and mobile parcel lockers, classify the attributes and divide each attribute into service attributes and scenario attributes. Collect user selection data for fixed and mobile parcel lockers for each attribute. Based on user selection data for fixed and mobile parcel lockers and user socio-demographic attribute data, construct a probability model for user selection of parcel lockers using multinomial Logit model and hybrid Logit model.

[0009] Step S3: Set different scenario attributes for fixed and mobile parcel lockers. Based on the probability model of users choosing parcel lockers, predict the adoption rate of mobile parcel lockers in different scenarios, and complete the prediction of users' demand for fixed and mobile parcel lockers.

[0010] As a preferred technical solution of the present invention: the preprocessing in step S1 includes filtering out the basic data of users who have no experience in using express delivery lockers.

[0011] As a preferred technical solution of the present invention: the user's socio-demographic attribute data in step S1 includes the user's gender, age, employment status, education level, income, and monthly online shopping frequency.

[0012] As a preferred technical solution of the present invention: in step S2, the service attributes of the fixed express cabinet include access distance, notification method, and daily delivery frequency of the fixed express cabinet; the service attributes of the mobile express cabinet include access distance, notification method, and dwell time of the mobile express cabinet; and the scenario attributes of the fixed and mobile express cabinets include package type, weather conditions, and date category.

[0013] As a preferred technical solution of the present invention, the specific steps of step S2 are as follows:

[0014] Step S21: Using the declarative experiment method, statistical analysis was conducted to determine users' preferences for fixed and mobile parcel lockers;

[0015] Step S22: Based on users' preferences for fixed and mobile parcel lockers, define the attributes of fixed and mobile parcel lockers respectively, and classify the attributes into service attributes and scenario attributes.

[0016] Step S23: The user chooses from three alternative options: fixed parcel locker, mobile parcel locker, and none of them; the user's selection data is collected.

[0017] Step S24: Construct a probabilistic model for users choosing parcel lockers using a multinomial Logit model and a hybrid Logit model. The specific method is as follows:

[0018] The utility of alternative i to user n in choosing task t nit It can be expressed as the following formula:

[0019] U nit =V nit +ε nit

[0020] In the formula, v nit ε represents deterministic utility. nit Represents random components;

[0021] Based on the influence of service attributes and context attributes, utility U nit Decomposed into the following formula:

[0022] U nit =β ni0 +x nit β ni +y n γ ni +ε nit

[0023] In the formula, x nit y is a service attribute variable of a vector. n Let β be a vector of socio-demographic attribute variables for user n. ni0 To replace a specific parameter, β ni and γ ni They are respectively with x nit and y n The relevant parameter vector;

[0024] Based on the hybrid Logit model, parameter β nik and substitute for specific parameter β ni0 It can be expressed as the following formula:

[0025] β nik =β ik +δ k v nik

[0026] β ni0 =β i0 +δ0v ni0

[0027] In the formula, β ik and β i0 δ is the population mean. k v nik and δ0v ni0 For additional error components, v nik and v ni0 Let δ be an independently distributed random term with a mean of 0 and a standard deviation of 1; k And δ0 are respectively β nik and β ni0 Standard deviation of the distribution;

[0028] The probability model P of user n choosing alternative i from T choice tasks ni As shown in the following formula:

[0029]

[0030] In the formula, β represents β ni and β nikAll information is given, j represents the alternative solutions other than the selected solution i, j ≠ i, and I is the set of all alternative solutions.

[0031] Beneficial effects: Compared with the prior art, the advantages of the present invention include:

[0032] 1. The present invention provides a method for predicting user demand for fixed and mobile parcel lockers, focusing the research on parcel lockers rather than introducing a large amount of research accumulated in the last-mile delivery field.

[0033] 2. This invention presents a method for predicting user demand for fixed and mobile parcel lockers, and studies the influence of contextual attributes, which have not been emphasized in previous studies.

[0034] 3. This invention presents a method for predicting user demand for fixed and mobile parcel lockers, comparing the two locker models from the customer's perspective. Previous studies have not taken this perspective. Attached Figure Description

[0035] Figure 1 This is a flowchart of a method for predicting user demand for fixed and mobile express delivery lockers according to an embodiment of the present invention;

[0036] Figure 2 These are actual images of fixed and mobile express delivery lockers provided according to embodiments of the present invention;

[0037] Figure 3 This is a socio-demographic attribute and percentage chart of users who selected "None" according to an embodiment of the present invention;

[0038] Figure 4 This is a schematic diagram of a user's selection set provided according to an embodiment of the present invention;

[0039] Figure 5 This is a diagram showing the estimation results of the model provided in the embodiment of the present invention;

[0040] Figure 6 This is a graph showing the change in user adoption rate of mobile parcel lockers according to an embodiment of the present invention. Detailed Implementation

[0041] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0042] Reference Figure 1 This invention provides a method for predicting user demand for fixed and mobile parcel lockers. For users in a target area, the following steps S1-S3 are performed to predict user demand for fixed and mobile parcel lockers:

[0043] Step S1: Collect socio-demographic attribute data of users in the target area and preprocess the socio-demographic attribute data of users. The preprocessing includes filtering out the basic data of users who have no experience in using express delivery lockers.

[0044] The following is a specific embodiment of the present invention:

[0045] The survey asked respondents to fill in data on their socio-demographic attributes, such as gender, age, employment status, education level, income, and monthly online shopping frequency. In this example, the survey data from Nanjing City from late May to June 2022 was used. After data cleaning, 5553 observations were generated for model estimation.

[0046] Descriptive statistics. Descriptive statistics of variables can provide a better understanding of the volatility in the data. The results of the sample descriptive statistics are shown in Table 1.

[0047] Table 1

[0048]

[0049] In Table 1, 51.7% of the respondents were female and 48.3% were male. Approximately one-third of the respondents were under 30 years old, while 27.9% were 50 years or older. Over half of the respondents indicated they were employed, and approximately 20% were students. Regarding education level, about 52.8% of respondents had a high school diploma or higher. In terms of monthly income, 30.3% of respondents earned less than 3,000 RMB, the average income was between 3,000 and 8,000 RMB, and about 16.9% reported a monthly income exceeding 8,000 RMB. Regarding the frequency of online shopping per month, 34% of respondents shopped online less than 3 times, while 27.6% shopped online 6 times or more per month.

[0050] Step S2: For both fixed and mobile parcel lockers, classify each attribute into service attributes and context attributes. Collect user selection data for each attribute for both fixed and mobile parcel lockers. Based on this data, along with user socio-demographic data, construct a probability model for user parcel locker selection using a multinomial Logit model and a hybrid Logit model. (See attached images of the fixed and mobile parcel lockers.) Figure 2 ,in Figure 2 (a) is a fixed parcel locker (SPL). Figure 2 (b) is a fixed parcel locker (MPL).

[0051] The service attributes of fixed parcel lockers include access distance, notification method, and daily delivery frequency. The service attributes of mobile parcel lockers include access distance, notification method, and dwell time. The contextual attributes of both fixed and mobile parcel lockers include parcel type, weather conditions, and date category.

[0052] The specific steps of step S2 are as follows:

[0053] Step S21: Use the declarative experiment method to statistically analyze users' preferences for fixed and mobile parcel lockers; the fixed and mobile parcel lockers targeted in the declarative experiment method can be based on real-world scenarios or hypothetical scenarios to quantify user preferences.

[0054] Step S22: Based on users' preferences for fixed and mobile parcel lockers, define the attributes of fixed and mobile parcel lockers respectively, and classify the attributes into service attributes and scenario attributes.

[0055] Detailed descriptions of service attributes and context attributes are shown in Table 2.

[0056] Table 2

[0057]

[0058] Step S23: Users choose from three options: fixed parcel lockers, mobile parcel lockers, and none of the above. User selection data is collected. The socio-demographic attributes and percentage of users who choose "none of the above" are detailed below. Figure 3 A diagram illustrating the selection set provided to the user based on three alternative options is shown below. Figure 4 .

[0059] Regarding the percentage of respondents choosing the "Neither" option (NC), male respondents were approximately 11.0%, slightly higher than female respondents. Regarding age, for respondents aged 50 and above who were unwilling to choose either of the two parcel locker options, the percentage was approximately 18.0%, while for respondents aged 30 to 49, the NC percentage dropped to 5.5%. Compared to employees and students, unemployed respondents had the highest percentage choosing the other 15 LMD (Location-Based Delivery) models. For both education level groups, the NC percentages were similar. Regarding monthly income, respondents earning less than 3,000 yuan per month had an NC percentage of 16.5%. Regarding online shopping frequency, respondents shopping less than 3 times per month had the highest NC percentage at 14.5%, while those shopping 6 times or more per month had only 4.7%.

[0060] Step S24: Construct a probabilistic model for users choosing parcel lockers using a multinomial Logit model and a hybrid Logit model. The specific method is as follows:

[0061] The utility of alternative i to user n in choosing task t nit It can be expressed as the following formula:

[0062] U nit =V nit +ε nit

[0063] In the formula, v nit ε represents deterministic utility. nit Represents a random component.

[0064] Based on the influence of service attributes and context attributes, utility U nit Decomposed into the following formula:

[0065] U nit =β ni0 +x nit β ni +y n γ ni +ε nit

[0066] In the formula, x nit y is a service attribute variable of a vector. n Let β be a vector of socio-demographic attribute variables for user n. ni0 To replace a specific parameter, β ni and γ ni They are respectively with x nit and y n The relevant parameter vector;

[0067] The attribute variables are coded hierarchically, with the first level coded as 1 and the second level as -1. For example, the sociodemographic attribute variable "gender" has two attribute levels: "female" is coded as 1, and "male" is coded as -1. For attribute variables with M levels (M≥3), there are M-1 index variables, each represented by a length vector 1×(M-1). For example, the sociodemographic attribute variable "age" has three attribute levels (less than 30, 30-49, greater than or equal to 50), which are coded as (1,0), (0,1), and (-1,-1), respectively. Other attribute variables are coded in the same way.

[0068] Considering the heterogeneity among users, a mixed Logit model is used to reflect the preference heterogeneity among different users and the potential panel effect caused by repeated selection experiments by the same user. The parameter β nik and substitute for specific parameter β ni0 It can be expressed as the following formula:

[0069] β nik =β ik +δ k vnik

[0070] β ni0 =β i0 +δ0v ni0

[0071] In the formula, β ik and β i0 δ is the population mean. k v nik and δ0v ni0 As an additional error component, v is used to represent the random effects caused by preference heterogeneity. nik and v ni0 Let δ be an independently distributed random term with a mean of 0 and a standard deviation of 1; k And δ0 are respectively β nik and β ni0 Standard deviation of the distribution;

[0072] The probability model P of user n choosing alternative i from T choice tasks ni As shown in the following formula:

[0073]

[0074] In the formula, β represents β ni and β nik All information is given, j represents the alternative solutions other than the selected solution i, j ≠ i, and I is the set of all alternative solutions.

[0075] In this embodiment, three models were used to investigate user preferences between two parcel locker modes: Model 1 is a basic MNL model, Model 2 is an ML model that incorporates latent preference heterogeneity among individuals, and Model 3 is a context-dependent ML model that considers preference heterogeneity, contextual attributes, and the interaction between service attributes and contextual attributes. The estimation results of the models are shown below. Figure 5 .

[0076] Step S3: Set different scenario attributes for fixed and mobile parcel lockers. Based on the probability model of users choosing parcel lockers, predict the adoption rate of mobile parcel lockers in different scenarios, and complete the prediction of users' demand for fixed and mobile parcel lockers.

[0077] In this embodiment, the adoption rate of mobile parcel lockers by users under different scenarios was simulated. A total of eight scenarios were generated based on three scenario attributes: parcel type, weather conditions, and date category. The changes in user adoption rate of mobile parcel lockers under different service attribute levels are shown in [the following text is incomplete and requires further context]. Figure 6Simulation results show that attribute levels representing high service quality for mobile parcel lockers can help improve adoption rates. Therefore, the setting of mobile parcel locker service attributes should be closely monitored during the initial introduction and implementation phases of the service.

[0078] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A method for predicting the demand of users of a fixed and mobile delivery cabinet, characterized by, For the target area users, the following steps S1-S3 are performed to complete the prediction of the users' demand for fixed and mobile express cabinets: Step S1: Collect the social demographic data of the users in the target area, and preprocess the social demographic data of the users; Step S2: For each attribute of the fixed and mobile express cabinets, respectively, attribute classification is performed, and each attribute is divided into service attributes and scenario attributes. The selection data of the users for each attribute of the fixed and mobile express cabinets is collected. According to the selection data of the users for the fixed and mobile express cabinets and the social demographic data of the users, a multi-item Logit model and a mixed Logit model are used to construct a probability model of the users' selection of express cabinets; The specific steps of step S2 are as follows: Step S21: Use the statement experiment method to count the users' preferences for the fixed and mobile express cabinets; Step S22: According to the users' preferences for the fixed and mobile express cabinets, define each attribute of the fixed and mobile express cabinets, respectively, and perform attribute classification. Each attribute is divided into service attributes and scenario attributes; Step S23: The users select from the fixed express cabinet, the mobile express cabinet, and none of the three alternatives. The selection data of the users is collected; Step S24: A multi-item Logit model and a mixed Logit model are used to construct a probability model of the users' selection of express cabinets. The specific method is as follows: Alternative i represents the utility of user n in selecting task t is represented by the following formula: ； wherein denotes the deterministic utility, denotes the stochastic component; According to the influence of service attributes and situational attributes, utility is decomposed into the following: ； wherein is a vector of service attribute variables, is a vector of social demographic attribute variables for user n, is a vector of alternative specific parameters, and are vectors of parameters related to and respectively. Based on the mixed Logit model, parameters and alternative specific constants are expressed as follows: ； ； where and are the overall means, and are additional error components, and are random terms with hypothetical independent distributions, mean 0, and standard deviation 1 ; and are the standard deviations of the and distributions, respectively; A probability model for user n to select alternative i in T selection tasks The following equation: ； In the formula, express and All information, In addition to the selected scheme Other alternatives, I is the set of all alternative solutions; Step S3: Set different scenario attributes of the fixed and mobile express cabinets. According to the probability model of the users' selection of express cabinets, predict the adoption rate of the mobile express cabinet in different scenarios, and complete the prediction of the users' demand for the fixed and mobile express cabinets. 2.The method of claim 1, wherein, The preprocessing in step S1 includes filtering out the basic data of users without express cabinet use experience. 3.The method of claim 1, wherein, The social demographic data of the users in step S1 includes the users' gender, age, employment status, education level, income, and monthly online shopping frequency. 4.The method of claim 1, wherein, In step S2, the service attributes of the fixed express cabinet include access distance, notification method, and daily delivery frequency of the fixed express cabinet. The service attributes of the mobile express cabinet include access distance, notification method, and stay time of the mobile express cabinet. The scenario attributes of the fixed and mobile express cabinets include package type, weather condition, and date category.

Images