Targeted marketing method and system for parking users

Abstract

The present invention belongs to the field of marketing and discloses a method and system for targeted marketing of parking users. The method comprises: S1, obtaining vehicle information and parking information of parking lot users; S2, performing data preprocessing on the parking information to obtain processed parking information; S3, performing data feature recognition on the processed parking information to obtain parking information features; S4, obtaining a user tag set based on the vehicle information and parking information features; S5, calculating a push value of an advertisement to be pushed based on the user tag set; and S6, determining an advertisement to be pushed to the parking lot user based on the push value. When basic attribute data and online data are unavailable, the accuracy of advertisement push in the present invention is effectively improved compared to randomly pushing advertisements.

Description

Technical Field

[0001] The present invention relates to the field of marketing, and in particular to a targeted marketing method and system for parking users. Background Art

[0002] Existing technologies typically build user profiles based on basic parking lot user attributes (e.g., gender, age, etc.) and online data (e.g., browsing history, purchase history, etc.), and then push advertisements based on these profiles. For example, patent application number CN201910260628.8 discloses similar technology. However, obtaining this data is often difficult because online data systems and parking lot management systems are often independent of each other, making it difficult for parking lot management systems to obtain this data. This results in insufficiently accurate marketing targeting parking lot users through parking lot management systems. Summary of the Invention

[0003] The purpose of the present invention is to disclose a targeted marketing method and system for parking users and to solve the technical problems pointed out in the background technology.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions:

[0005] In one aspect, the present invention provides a targeted marketing method for parking users, comprising:

[0006] S1, obtain the parking lot user's vehicle information and parking information;

[0007] S2, performing data preprocessing on the parking information to obtain processed parking information;

[0008] S3, performing data feature recognition on the processed parking information to obtain parking information features;

[0009] S4, obtaining a user tag set based on vehicle information and parking information features;

[0010] S5, calculating the push value of the advertisement to be pushed based on the user tag set;

[0011] S6: Determine an advertisement to be pushed to parking lot users based on the push value.

[0012] Furthermore, the vehicle information includes make and model;

[0013] Parking information includes parking location and parking duration.

[0014] Furthermore, the process of obtaining parking information includes:

[0015] Get the parking information of the N time intervals closest to the current time, where N is the adaptive control coefficient.

[0016] Furthermore, the process of determining the N time intervals includes:

[0017] The current time is expressed as tn, and the Nth time interval is ,T is the length of the set time interval;

[0018] The nth time interval is , n∈[1,N-1].

[0019] Furthermore, the parking information is pre-processed to obtain processed parking information, including:

[0020] Perform data cleaning on the parking information to obtain processed parking information.

[0021] Furthermore, data feature recognition is performed on the processed parking information to obtain parking information features, including:

[0022] The first step is to obtain the characteristics of the parking locations included in the parking information as follows:

[0023] Get the set of parking locations in each time interval respectively;

[0024] Calculate the cluster center of each parking location set separately;

[0025] Calculate activity range based on cluster centers;

[0026] In the second step, the feature acquisition process for the parking duration included in the parking information is as follows:

[0027] Calculate the impact coefficient for each time interval separately;

[0028] Calculate the average parking time of each parking location in each time interval;

[0029] Calculate the corrected parking time of each parking location in the Nth time interval based on the influence coefficient and the average value of the parking time;

[0030] In the third step, the activity range and the corrected parking duration are used as parking feature information.

[0031] Furthermore, a user tag set is obtained based on the vehicle information and parking information features, including:

[0032] Acquire a first tag set according to vehicle information;

[0033] Acquire a second tag set according to parking information features;

[0034] The union of the first tag set and the second tag set is used as the user tag set.

[0035] Furthermore, the push value of the advertisement to be pushed is calculated based on the user tag set, including:

[0036] Get the ad tag set of the ad to be pushed;

[0037] The push value of the advertisement to be pushed is calculated based on the user tag set and the advertisement tag set.

[0038] Furthermore, determining an advertisement to be pushed to parking lot users based on the push value includes:

[0039] The advertisement to be pushed corresponding to the largest push value is used as the advertisement pushed to the parking lot users.

[0040] On the other hand, the present invention provides a targeted marketing system for parking users, including an information acquisition module, a pre-processing module, a feature recognition module, a tag acquisition module, a calculation module, and a determination module;

[0041] The information acquisition module is used to obtain the vehicle information and parking information of parking lot users;

[0042] The preprocessing module is used to perform data preprocessing on the parking information to obtain processed parking information;

[0043] The feature recognition module is used to perform data feature recognition on the processed parking information to obtain parking information features;

[0044] The tag acquisition module is used to obtain a user tag set based on vehicle information and parking information features;

[0045] The calculation module is used to calculate the push value of the advertisement to be pushed based on the user tag set;

[0046] The determination module is used to determine the advertisement to be pushed to the parking lot user based on the push value.

[0047] Beneficial effects:

[0048] This invention doesn't build user profiles based on conventional basic attribute data and online data. Instead, it utilizes parking management systems to obtain vehicle and parking information, which is then used to select ads for push notifications. Vehicle make and model directly reflect a user's spending power, while parking lot attributes (such as high-end residential complexes and schools) can infer a user's lifestyle and needs. When basic attribute data and online data are unavailable, this invention significantly improves the accuracy of ad push notifications compared to random ad push notifications. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0050] Figure 1 Schematic diagram of the targeted marketing method for parking users of the present invention.

[0051] Figure 2 Schematic diagram of the targeted marketing system for parking users of the present invention. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0053] like Figure 1 In one embodiment shown, the present invention provides a targeted marketing method for parking users, comprising:

[0054] S1, obtain the vehicle information and parking information of the parking lot user.

[0055] Furthermore, the vehicle information includes make and model;

[0056] Parking information includes parking location and parking duration.

[0057] Specifically, the vehicle may be photographed when entering or exiting a parking lot gate to obtain a vehicle appearance image (eg, a vehicle rear image), and then the vehicle appearance image may be recognized to obtain vehicle information.

[0058] For example, vehicle information can be:

[0059] Brand: Baox;

[0060] Model: xxxLi.

[0061] Parking information can be obtained through the database of the parking lot management system. When multiple parking lots are managed by the same property management company, the multiple parking lots use the same management system, so that the present invention can conduct targeted marketing to parking lot users based on the parking information of multiple parking lots.

[0062] In addition, the authorization of parking information can be obtained from property management companies that manage other parking lots to expand the coverage of the parking lot of the present invention, thereby obtaining more accurate marketing results.

[0063] Furthermore, the parking location is the location of the parking lot, not the location of the vehicle.

[0064] Furthermore, the process of obtaining parking information includes:

[0065] Get the parking information of the N time intervals closest to the current time, where N is the adaptive control coefficient.

[0066] In the present invention, the value of N is not fixed, but can change with the actual situation of parking lot users.

[0067] When a user drives out of a parking lot, parking information is generated and stored in a database. Therefore, based on the storage time, the time interval into which each piece of parking information falls can be determined.

[0068] Specifically, the process of obtaining N includes:

[0069] Get the first two time intervals L1 and L2 closest to the current time;

[0070] The parking information C1 and C1 of L1 and L2 are obtained respectively;

[0071] Obtain the sets P1 and P2 of parking locations in C1 and C2 respectively, and obtain the sets TM1 and TM2 of parking durations in C1 and C2 respectively;

[0072] Calculate the coefficient of variation based on P1, P2, TM1, and TM2;

[0073] N is calculated based on the coefficient of variation.

[0074] When calculating N, the present invention uses the two most recent time intervals from the current time. L1 can be one of the two most recent time intervals, while L2 is the other. By acquiring a set of parking locations, it is possible to determine whether there have been significant changes in parking locations in the two most recent time intervals. Furthermore, by acquiring a set of parking times, it is possible to compare whether the parking duration at the same parking location has significantly changed in these two time intervals. This allows for a more accurate determination of the value of N, reducing the computational complexity of subsequent feature recognition. This allows the present invention to maintain the efficiency of obtaining ads pushed to each parking user even when there are many parking lot users to whom ads need to be pushed.

[0075] Furthermore, the coefficient of variation is calculated based on P1, P2, TM1, and TM2, including:

[0076] Get the intersection P3 of P1 and P2;

[0077] The coefficient of variation is calculated using the following function:

[0078]

[0079] represents the coefficient of variation, 、 and Represent the total number of parking places included in P1, P2 and P3 respectively, and denote the average parking duration of parking location i in time interval L1 and time interval L2, respectively. Indicates weight, max means getting the larger value in the brackets, Represents the median of the set tlen, tlen is i∈P3 The set of calculation results.

[0080] For example, the value of max(1,2) is 2.

[0081] The above average value is the average of all parking durations at the same parking location in the same time period.

[0082] The variation coefficient of the present invention is calculated from the number of identical parking locations in two time intervals and the average parking duration at the identical parking location in the two time intervals. Therefore, the greater the number of identical parking locations in the two time intervals and the less the variation in the average parking duration at the identical parking location in the two time intervals, the more stable the living habits of the parking lot users are. At this time, sufficiently accurate parking information features can be obtained by relying only on a small amount of historical parking information, thereby improving the efficiency of subsequent acquisition of parking information features.

[0083] Furthermore, the value of the weight may be 0.5.

[0084] Furthermore, N is calculated based on the coefficient of variation, including:

[0085] Use the following function to calculate N:

[0086]

[0087] Indicates the preset control value, PN indicates the preset quantity, is the rounding symbol, for example, The value of is 3.

[0088] In the present invention, the value of N changes with the change of the coefficient of variation. When the coefficient of variation is larger, it means that the user's parking habits have changed more. Therefore, more historical parking information needs to be introduced to calculate the parking feature information, so as to reduce the interference degree of sudden changes in parking behavior as noise on the parking feature information. Therefore, a more accurate label set can be obtained.

[0089] Furthermore, PN may be 10.

[0090] Further, The value of can be 0.6.

[0091] Furthermore, the process of determining the N time intervals includes:

[0092] The current time is expressed as tn, and the Nth time interval is ,T is the length of the set time interval;

[0093] The nth time interval is , n∈[1,N-1].

[0094] Furthermore, T may be 1 week.

[0095] Furthermore, the current time may be the time when the parking lot user last left the parking lot.

[0096] S2: Perform data preprocessing on the parking information to obtain processed parking information.

[0097] Furthermore, the parking information is pre-processed to obtain processed parking information, including:

[0098] Perform data cleaning on the parking information to obtain processed parking information.

[0099] Data cleaning is the process of identifying and correcting errors, inconsistencies, duplications, or invalid information in data through technical means, aiming to improve data accuracy, completeness, and consistency. Core tasks include addressing missing values, correcting erroneous data, eliminating duplicate records, standardizing data formats, and handling outliers.

[0100] Cleaning of parking lots:

[0101] Address standardization:

[0102] Convert parking location text (e.g., "xx Road, xxx") to latitude and longitude coordinates or a unified administrative division format for subsequent map visualization or spatial analysis. For example, use a geocoding tool (such as the Python geocoder library) to convert the text address to latitude and longitude.

[0103] Wrong address correction:

[0104] Use external data sources (such as map APIs) to verify address validity and correct spelling errors or ambiguous expressions (for example, "a shopping mall parking lot" requires a specific name).

[0105] Missing value handling:

[0106] If the parking location is missing, it can be supplemented by associating it with other data sources (such as parking management system logs) based on the vehicle entry and exit time; if it cannot be supplemented, it will be marked as "unknown" or the record will be deleted.

[0107] Cleaning during parking time:

[0108] Outlier Detection:

[0109] The logical range check parking time must be greater than 0 and less than a reasonable upper limit (such as 72 hours). Exceeding the range is considered an abnormality.

[0110] Statistical Identification: Detect outliers using box plots or Z-scores. For example, parking durations with Z-scores exceeding ±3 may indicate erroneous data.

[0111] Error duration correction:

[0112] If the parking duration is abnormal (such as a negative value) due to equipment failure, it must be corrected based on the records of the adjacent time period or manual review.

[0113] Missing duration processing:

[0114] When using interpolation methods (such as filling with the mean of previous and subsequent records) or directly deleting missing records, you need to balance data completeness and accuracy based on the business scenario.

[0115] S3, performing data feature recognition on the processed parking information to obtain parking information features.

[0116] This step is to extract features by integrating historical parking information, so that the obtained parking information features can not only include the parking features of the current time, but also the parking features of the past. This can reduce the impact of the parking lot users' accidental parking behaviors that are significantly different from their parking habits on the accuracy of the parking lot users' label judgment.

[0117] Furthermore, data feature recognition is performed on the processed parking information to obtain parking information features, including:

[0118] The first step is to obtain the characteristics of the parking locations included in the parking information as follows:

[0119] Get the set of parking locations in each time interval respectively;

[0120] Calculate the cluster center of each parking location set separately;

[0121] Calculate activity range based on cluster centers;

[0122] In the second step, the feature acquisition process for the parking duration included in the parking information is as follows:

[0123] Calculate the impact coefficient for each time interval separately;

[0124] Calculate the average parking time of each parking location in each time interval;

[0125] Calculate the corrected parking time of each parking location in the Nth time interval based on the influence coefficient and the average value of the parking time;

[0126] In the third step, the activity range and the corrected parking duration are used as parking feature information.

[0127] The present invention acquires features based on the parking location and parking duration respectively, thereby making the parking information features more comprehensive.

[0128] Specifically, by determining the activity range, in the subsequent process, advertisements of merchants near the activity range can be matched based on the activity range, thereby achieving more accurate advertising push.

[0129] By calculating and correcting the parking time, the impact of occasional deviations from parking habits on the label determination process can be reduced, further improving the accuracy of the label.

[0130] Furthermore, the cluster centers of each parking location set are calculated separately, including:

[0131] Using the clustering algorithm, the number of cluster centers is set to 1, and the cluster center of each set of parking places is calculated respectively.

[0132] For example, an algorithm such as K-means clustering can be used to obtain cluster centers.

[0133] Furthermore, the activity range is calculated based on the cluster center, including:

[0134] The range with a radius of R and the cluster center as the center is used as the activity range.

[0135] In the present invention, the value of R may be 5 kilometers.

[0136] Furthermore, the impact coefficient of each time interval is calculated separately, including:

[0137] Use the following function to calculate the influence coefficient of the time interval:

[0138]

[0139] Represents the influence coefficient of the mth time interval. When m∈[1,N-1], Indicates the start time of the mth time interval. When m is equal to N, Indicates the end time of the N-1th time interval; Indicates the start time of the first time interval. represents the set of parking places included in the mth time interval, tn is the current time, express The standard deviation of parking time at parking location j in is:

[0140]

[0141] represents the total number of parking times at parking location j in the mth time interval, represents the parking duration of the kth stop in the mth time interval;

[0142] represents the total number of parking places included in the mth time interval, represents the median of the standard deviation of parking durations at all parking locations included in the mth time interval; is the time impact weight.

[0143] In calculating the influence coefficient, the present invention incorporates the time interval between the mth time interval and the first time interval, as well as the sum of the standard deviations of parking durations at each parking location within the mth time interval. This allows the influence coefficient to be calculated by integrating two different types of data, resulting in a more accurate calculated influence coefficient. The longer the time interval between the mth time interval and the first time interval, and the smaller the sum of the standard deviations of parking durations at each parking location, the greater the influence coefficient, indicating that the data from the mth time interval serves as a greater reference for subsequent calculations and corrections to parking durations. This allows historical parking information to be incorporated into the calculation of parking information features, while also avoiding retaining the same reference for parking durations across all time intervals, enabling the present invention to more promptly adapt to changes in parking durations.

[0144] Furthermore, the time impact weight is 0.6.

[0145] Furthermore, based on the influence coefficient and the average value of the parking time, the corrected parking time of each parking location in the Nth time interval is calculated, including:

[0146] Store all parking places in the Nth time interval into a set ;

[0147] for The zth parking spot in , The calculation formula for the corrected parking time is:

[0148]

[0149] express Corrected parking time, is the influence coefficient of the qth time interval, is the parking location in the qth time interval The average parking time.

[0150] The corrected parking time is calculated based on the influence coefficient. For the same parking location, the larger the influence coefficient of the time interval corresponding to the parking time, the greater the impact of the parking time on the final corrected parking time. In this way, different nonlinear influence levels are set for different time intervals, so that the final corrected parking time can more accurately reflect the parking habits of parking lot users.

[0151] S4, obtaining a user tag set based on vehicle information and parking information features.

[0152] In this step, the user tags of parking lot users are mainly obtained based on the pre-set tag classification standards. Furthermore, the user tag set is obtained based on the vehicle information and parking information features, including:

[0153] Acquire a first tag set according to vehicle information;

[0154] Acquire a second tag set according to parking information features;

[0155] The union of the first tag set and the second tag set is used as the user tag set. Further, obtaining the first tag set based on the vehicle information includes:

[0156] Obtain the price of the vehicle in the used car market based on vehicle information;

[0157] Get user tags based on price;

[0158] Get the type of vehicle based on vehicle information;

[0159] Get user tags based on vehicle type;

[0160] The user tag obtained based on the price and the user tag obtained based on the vehicle type are stored in the first tag set.

[0161] For example, if the price is less than 200,000, the user label is an entry-level car owner; if the price is greater than or equal to 200,000 and less than 500,000, the user label is a mid-range car owner; if the price is greater than or equal to 500,000, the user label is a luxury car owner.

[0162] The prices in the used car market can be obtained from platforms such as xxdi and xxzhijia.

[0163] The types of vehicles include sedans, SUVs, MPVs, etc. The user tags corresponding to sedans, SUVs, and MPVs can be sedan owners, SUV owners, and MPV owners, respectively.

[0164] In addition, in addition to vehicle price and vehicle type, user labels can also be obtained based on data such as vehicle color.

[0165] Furthermore, obtaining a second tag set based on parking information features includes:

[0166] User tags are obtained according to the activity range and the corrected parking duration, and the obtained user tags are stored in the second tag set.

[0167] Specifically, when obtaining user labels based on the activity range, the average housing price in the activity range can be used as the judgment standard for user labels. For example, when the average housing price is greater than or equal to 100,000, the user label is a CBD person; when the average housing price is less than 100,000 and greater than or equal to 70,000, the user label is a school district housing person; when the average housing price is less than 70,000 and greater than or equal to 40,000, the user label is a person in the emerging industrial-city integration zone; and for other average housing prices, the user label is a person in the suburban residential area.

[0168] When obtaining user labels based on the corrected parking duration, first calculate the average corrected parking duration of each parking location in the Nth time interval;

[0169] Set corresponding location weights for each parking location based on the area type to which the parking location belongs;

[0170] For example, the regional types are divided into CBD, school district housing, emerging industrial-city integration areas and suburban residential areas, and the location weights are set to 0.65, 0.2, 0.1 and 0.05 respectively.

[0171] The average value of the corrected parking time at different parking locations is weighted and summed according to the location weight to obtain the parking time judgment value:

[0172]

[0173] Indicates the parking time judgment value, is the location weight of the c-th parking location in the N-th time interval, is the normalized value corresponding to the average value of the corrected parking duration at the c-th parking location; cN represents the set of parking locations in the N-th time interval;

[0174] The user tag is obtained based on the parking duration judgment value.

[0175] For example, the relationship between the parking duration judgment value and the user label can be:

[0176] If the parking duration judgment value is greater than or equal to 0.8, the user is labeled as a high-spending group;

[0177] If the parking duration judgment value is greater than or equal to 0.4 and less than 0.8, the user is labeled as a medium-spending group;

[0178] The parking time judgment value is less than 0.4, and the user label is a low-spending ability group.

[0179] S5: Calculate the push value of the advertisement to be pushed based on the user tag set.

[0180] In this step, the push value is calculated mainly by comparing the correlation between the tags in the user tag set and the advertisement to be pushed.

[0181] Furthermore, the push value of the advertisement to be pushed is calculated based on the user tag set, including:

[0182] Get the ad tag set of the ad to be pushed;

[0183] The push value of the advertisement to be pushed is calculated based on the user tag set and the advertisement tag set.

[0184] In the present invention, the advertising tags of the advertisements to be pushed are tags that are manually set in advance and stored in the database, and the advertising tags of the advertisements to be pushed can be obtained from the database; for example, for high-end brand advertisements, the advertising tags may include luxury car owners, CBD personnel, high-spending groups, etc.

[0185] Furthermore, the push value of the advertisement to be pushed is calculated based on the user tag set and the advertisement tag set, including:

[0186] Use the following function to calculate the push value of the ad to be pushed:

[0187]

[0188] express and The total number of labels contained in the union of ; express The total number of tags contained in , and are the user tag set and the ad tag set of the ad b to be pushed, The push value of advertisement b to be pushed.

[0189] In another embodiment, calculating the push value of an advertisement to be pushed based on a user tag set includes:

[0190] S6: Determine an advertisement to be pushed to parking lot users based on the push value.

[0191] This step can push the ads based on the push value. Generally speaking, ads can be pushed to mini programs, apps, and other terminals that can install the parking management system client, and pop-up push can be performed when the user opens the client.

[0192] Furthermore, determining an advertisement to be pushed to parking lot users based on the push value includes:

[0193] The advertisement to be pushed corresponding to the largest push value is used as the advertisement pushed to the parking lot users.

[0194] On the other hand, Figure 2 ,The present invention provides a targeted marketing system for parking users, including an information acquisition module, a pre-processing module, a feature recognition module, a label acquisition module, a calculation module and a determination module;

[0195] The information acquisition module is used to obtain the vehicle information and parking information of parking lot users;

[0196] The preprocessing module is used to perform data preprocessing on the parking information to obtain processed parking information;

[0197] The feature recognition module is used to perform data feature recognition on the processed parking information to obtain parking information features;

[0198] The tag acquisition module is used to obtain a user tag set based on vehicle information and parking information features;

[0199] The calculation module is used to calculate the push value of the advertisement to be pushed based on the user tag set;

[0200] The determination module is used to determine the advertisement to be pushed to the parking lot user based on the push value.

[0201] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A targeted marketing method for parking users, characterized in that: include: S1, obtain the parking lot user's vehicle information and parking information; S2, performing data preprocessing on the parking information to obtain processed parking information; S3, performing data feature recognition on the processed parking information to obtain parking information features, including: The first step is to obtain the characteristics of the parking locations included in the parking information as follows: Get the set of parking locations in each time interval respectively; Calculate the cluster center of each parking location set separately; Calculate activity range based on cluster centers; In the second step, the feature acquisition process for the parking duration included in the parking information is as follows: Calculate the impact coefficient for each time interval separately, including: Use the following function to calculate the influence coefficient of the time interval: ； Represents the influence coefficient of the mth time interval. When m∈[1,N-1], Indicates the start time of the mth time interval. When m is equal to N, Indicates the end time of the N-1th time interval; Indicates the start time of the first time interval. represents the set of parking places included in the mth time interval, tn is the current time, express The standard deviation of parking time at parking location j in is: ； represents the total number of parking times at parking location j in the mth time interval, represents the parking duration of the kth stop in the mth time interval; represents the total number of parking places included in the mth time interval, represents the median of the standard deviation of parking durations at all parking locations included in the mth time interval; is the time impact weight; If the time interval between the mth time interval and the first time interval is longer, and the sum of the standard deviations of the parking durations at each parking location is smaller, then the influence coefficient is larger, indicating that the data of the mth time interval has a greater reference value for the subsequent calculation and correction of the parking duration. Calculate the average parking time of each parking location in each time interval; Calculate the corrected parking time of each parking location in the Nth time interval based on the influence coefficient and the average value of the parking time; The third step is to use the activity range and the corrected parking duration as parking feature information; S4, obtaining a user tag set based on vehicle information and parking information features; S5, calculating the push value of the advertisement to be pushed based on the user tag set; S6, determining an advertisement to be pushed to parking lot users based on the push value; Vehicle information including make and model; Parking information includes parking location and parking duration; The process of obtaining parking information includes: Get parking information for the N time intervals closest to the current time; The process of obtaining N includes: Get the first two time intervals L1 and L2 closest to the current time; The parking information C1 and C1 of L1 and L2 are obtained respectively; Obtain the sets P1 and P2 of parking locations in C1 and C2 respectively, and obtain the sets TM1 and TM2 of parking durations in C1 and C2 respectively; Calculate the coefficient of variation based on P1, P2, TM1, and TM2; N is calculated based on the variation coefficient. The value of N changes with the variation coefficient. When the variation coefficient is larger, it means that the user's parking habits have changed more significantly, and more historical parking information is needed to calculate the parking feature information. Calculates the coefficient of variation based on P1, P2, TM1, and TM2, including: Get the intersection P3 of P1 and P2; The coefficient of variation is calculated using the following function: ； represents the coefficient of variation, 、 and Represent the total number of parking places included in P1, P2 and P3 respectively, and denote the average parking duration of parking location i in time interval L1 and time interval L2, respectively. Indicates weight, max means getting the larger value in the brackets, Represents the median of the set tlen, tlen is i∈P3 The set of calculation results.

2. The targeted marketing method for parking users according to claim 1, characterized in that: The process of determining N time intervals includes: The current time is expressed as tn, and the Nth time interval is ,T is the length of the set time interval; The nth time interval is , n∈[1,N-1].

3. The targeted marketing method for parking users according to claim 2, characterized in that: Perform data preprocessing on parking information to obtain processed parking information, including: Perform data cleaning on the parking information to obtain processed parking information.

4. The targeted marketing method for parking users according to claim 1, characterized in that: Obtain a user tag set based on vehicle information and parking information features, including: Acquire a first tag set according to vehicle information; Acquire a second tag set according to parking information features; The union of the first tag set and the second tag set is used as the user tag set.

5. The targeted marketing method for parking users according to claim 1, characterized in that: Calculate the push value of the advertisement to be pushed based on the user tag set, including: Get the ad tag set of the ad to be pushed; The push value of the advertisement to be pushed is calculated based on the user tag set and the advertisement tag set.

6. The targeted marketing method for parking users according to claim 1, characterized in that: Advertisements pushed to parking lot users are determined based on push values, including: The advertisement to be pushed corresponding to the largest push value is used as the advertisement pushed to the parking lot users.

7. Targeted marketing system for parking users, characterized by: It includes information acquisition module, preprocessing module, feature recognition module, label acquisition module, calculation module and determination module; The information acquisition module is used to obtain the vehicle information and parking information of parking lot users; The preprocessing module is used to perform data preprocessing on the parking information to obtain processed parking information; The feature recognition module is used to perform data feature recognition on the processed parking information to obtain parking information features, including: The first step is to obtain the characteristics of the parking locations included in the parking information as follows: Get the set of parking locations in each time interval respectively; Calculate the cluster center of each parking location set separately; Calculate activity range based on cluster centers; In the second step, the feature acquisition process for the parking duration included in the parking information is as follows: Calculate the impact coefficient for each time interval separately, including: Use the following function to calculate the influence coefficient of the time interval: ； Represents the influence coefficient of the mth time interval. When m∈[1,N-1], Indicates the start time of the mth time interval. When m is equal to N, Indicates the end time of the N-1th time interval; Indicates the start time of the first time interval. represents the set of parking places included in the mth time interval, tn is the current time, express The standard deviation of parking time at parking location j in is: ； represents the total number of parking times at parking location j in the mth time interval, represents the parking duration of the kth stop in the mth time interval; represents the total number of parking places included in the mth time interval, represents the median of the standard deviation of parking durations at all parking locations included in the mth time interval; is the time impact weight; If the time interval between the mth time interval and the first time interval is longer, and the sum of the standard deviations of the parking durations at each parking location is smaller, then the influence coefficient is larger, indicating that the data of the mth time interval has a greater reference value for the subsequent calculation and correction of the parking duration. Calculate the average parking time of each parking location in each time interval; Calculate the corrected parking time of each parking location in the Nth time interval based on the influence coefficient and the average value of the parking time; The third step is to use the activity range and the corrected parking duration as parking feature information; The tag acquisition module is used to obtain a user tag set based on vehicle information and parking information features; The calculation module is used to calculate the push value of the advertisement to be pushed based on the user tag set; The determination module is used to determine the advertisement to be pushed to the parking lot user based on the push value; Vehicle information including make and model; Parking information includes parking location and parking duration; The process of obtaining parking information includes: Get parking information for the N time intervals closest to the current time; The process of obtaining N includes: Get the first two time intervals L1 and L2 closest to the current time; The parking information C1 and C1 of L1 and L2 are obtained respectively; Obtain the sets P1 and P2 of parking locations in C1 and C2 respectively, and obtain the sets TM1 and TM2 of parking durations in C1 and C2 respectively; Calculate the coefficient of variation based on P1, P2, TM1, and TM2; N is calculated based on the variation coefficient. The value of N changes with the variation coefficient. When the variation coefficient is larger, it means that the user's parking habits have changed more significantly, and more historical parking information is needed to calculate the parking feature information. Calculates the coefficient of variation based on P1, P2, TM1, and TM2, including: Get the intersection P3 of P1 and P2; The coefficient of variation is calculated using the following function: ； represents the coefficient of variation, 、 and Represent the total number of parking places included in P1, P2 and P3 respectively, and denote the average parking duration of parking location i in time interval L1 and time interval L2, respectively. Indicates weight, max means getting the larger value in the brackets, Represents the median of the set tlen, tlen is i∈P3 The set of calculation results.

Images