A vehicle heat generation method based on historical data

Abstract

This invention relates to the fields of vehicle data processing and vehicle operation management technology, and discloses a method for generating vehicle popularity based on historical data. The method establishes a vehicle popularity generation session, reads vehicle behavior, conversion, price, inventory, exposure, and event history records; aggregates data according to stages such as front-end contact, interest retention, intention submission, in-store test drive, transaction formation, transaction withdrawal, and abnormal attention; registers the exposure impact status, event noise status, price status, and inventory status; connects natural behavior records, back-end conversion records, and late return records according to the future demand observation cycle, and isolates records of strong exposure without conversion, negative event attention records, and records with abnormal sources; and generates a vehicle popularity version based on the evidence chain of real vehicle demand. This method can reduce the interference of platform exposure, short-term events, and conversion lag on vehicle popularity.

Description

Technical Field

[0001] This invention relates to the fields of vehicle data processing and vehicle operation management technology, specifically a method for generating vehicle heat maps based on historical data. Background Technology

[0002] Vehicle trading platforms, automotive information platforms, rental vehicle allocation platforms, and dealer management systems typically need to generate vehicle popularity scores based on historical data when recommending vehicles, allocating inventory, distributing sales resources, and assessing regional demand. Existing methods often statistically analyze, weight, or aggregate data such as page views, clicks, searches, favorites, model comparisons, inquiry leads, test drive appointments, transaction records, price changes, inventory records, and news / public opinion over time windows to obtain a popularity score for a vehicle, model, or regional combination of vehicles.

[0003] However, the formation mechanisms and implications of the aforementioned historical data differ. Browsing, clicking, searching, and saving are low-cost front-end behaviors, easily influenced by recommendation slots, advertising, special event zones, and platform ranking strategies; inquiry leads, test drive appointments, and transaction records are closer to actual demand, but usually lag behind browsing, searching, and comparison behaviors; price reductions, inventory changes, subsidy policies, competitor product launches, media reviews, quality complaints, and recall events also cause fluctuations in attention data within different time windows. Directly merging the above data into the same statistical window can easily treat data from different feedback periods, sources, and noise types as evidence of similar popularity.

[0004] In real-world scenarios, after a vehicle is recommended by a platform or advertised, its page views and clicks may increase in the short term, but inquiries, test drives, and sales may not increase accordingly. When a model receives attention due to complaints, recalls, or controversial news, its search volume and comments may also rise, but this increase does not necessarily represent positive demand. Some vehicles with price adjustments or regional inventory shortages may have low initial page views, but they may generate continuous leads and sales in subsequent periods. Existing methods for generating vehicle popularity struggle to distinguish between short-term noise-driven popularity and convertible, sustainable popularity, and also struggle to eliminate the cyclical amplification effect of platform exposure. This leads to the generated vehicle popularity value easily misjudging advertising traffic, short-term attention, or negative public opinion as genuine demand popularity, and may underestimate the sustained demand popularity with delayed sales feedback, thus affecting the accuracy of vehicle recommendations, inventory allocation, regional deployment, and sales follow-up. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for generating vehicle heat maps based on historical data, thus solving the aforementioned problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for generating vehicle heat maps based on historical data, comprising:

[0007] S1. Establish a vehicle popularity generation session, register the vehicle object number, historical observation time range, future demand observation cycle, region number, channel number, data cutoff time and popularity configuration version number, and read vehicle behavior, conversion, price, inventory, exposure and event history.

[0008] S2. Organize historical records based on vehicle object number, behavior occurrence time, data reception time, channel number, exposure batch number, price version number, and inventory version number, and categorize them according to the stages of front-end contact, interest retention, intention submission, in-store test drive, transaction formation, transaction withdrawal, and abnormal attention.

[0009] S3. Based on the exposure batch, entry type, page location, activity number, event type, price version, and inventory version, register the exposure impact status, event noise status, price status, and inventory status for the historical records of each stage.

[0010] S4. According to the phase response window corresponding to the future demand observation cycle, connect the natural behavior records, backend conversion records and late return records of the same vehicle object, and isolate the strong exposure non-conversion records, negative event attention records and abnormal source records.

[0011] S5. Generate a vehicle popularity version based on the evidence chain of real vehicle demand after connection and its corresponding status, and register the popularity level, popularity direction, popularity credibility status, popularity validity period and version output range.

[0012] Furthermore, the vehicle object number is formed by combining the unique vehicle listing number, the de-identified vehicle identification number, the sales store number, and the listing channel number in a fixed order under the single vehicle for sale category; and by combining the brand number, model number, model year number, configuration number, energy type number, region number, and channel number in a fixed order under the vehicle configuration combination category.

[0013] When a field is missing, write a missing placeholder at the corresponding concatenation position and register the missing field status of the object.

[0014] Furthermore, during the historical record organization, inquiries, telephone consultations, online customer service, quotation requests, and contact information are grouped into consultation leads; test drive appointments, in-store appointments, and test drive registrations are grouped into test drive appointments; deposit payments, intention deposit payments, and order lock records are grouped into down payments; and contract signing, invoicing, and delivery registration are grouped into transaction formation records.

[0015] The aggregation criteria are that at least two of the original behavior name, entry type, business result field, and subsequent business records point to the same behavior type.

[0016] Furthermore, after phased aggregation, duplicate records of the same vehicle object number, the same user de-identified number, the same device de-identified number, and the same behavior type are merged.

[0017] The initial contact phase uses a 30-minute repeat window and retains the first valid contact record; the interest retention phase uses a 24-hour repeat window and retains the last valid status; and the intention submission phase uses a 7-day repeat window and retains the first valid inquiry lead.

[0018] Furthermore, the exposure impact status is registered based on the exposure batch number, entry type, page location, activity number, exposure start time, and exposure end time;

[0019] When the page position is an advertisement position, a homepage focus position, a first screen position in the information feed, a splash screen position, or a recommendation position, the corresponding front-end contact stage is recorded as a high-exposure contact record.

[0020] When an activity number corresponds to a lottery, red envelope, points, task rewards, or new user acquisition promotion, the corresponding front-end contact stage record is registered as an activity incentive contact record.

[0021] Furthermore, the event noise status is registered based on the event type, event occurrence time, event release channel, affected vehicle objects, and event direction confirmation fields;

[0022] Quality complaints, recalls, negative reviews, delivery delays, and controversial news are registered as negative attention events; official price reductions, financial subsidies, trade-in subsidies, and limited-time offers are registered as transaction stimulus events; competitor product launches, competitor price reductions, and competitor product recalls are registered as competitor product disturbance events; and tight inventory, extended delivery cycles, and shortages of popular configurations are registered as supply restriction events.

[0023] Furthermore, the price status is registered based on the price version number, official guide price, platform display price, store price, discount amount, financial subsidy, trade-in subsidy, and price effective time;

[0024] When the platform display price, store price, and discount amount are in the same direction in two consecutive price versions and the change does not reach the registration threshold of the popularity configuration version, the registered price is in a stable state. When the platform display price decreases, the store price decreases, the discount amount increases, and the subsidy field takes effect, the registered price is in a downward state. When there is a conflict in the direction of increase or decrease among the official guide price, platform display price, and store price, the registered price is in a chaotic state.

[0025] Furthermore, inventory status is registered based on inventory version number, available quantity, quantity in transit, locked order quantity, sold but not delivered quantity, delisted quantity, inventory update time, and store availability flag.

[0026] When the available quantity reaches the registration threshold for the popularity configuration version and the number of locked orders does not reach the threshold, the inventory status is registered as sufficient. When the available quantity is lower than the threshold, the number of locked orders increases, and there are sold but undelivered quantities, the inventory status is registered as tight. When at least one of the following exists, the inventory status is registered as frozen: store frozen, procedures pending, unsaleable, or delisted.

[0027] Furthermore, the phased response window is set according to the future demand observation cycle;

[0028] The front-end contact stage records a valid response when the device connects to favorites, model comparisons, and price subscriptions within the first response window. The interest retention stage records an interest-to-intent response when the device connects to inquiry leads, quote requests, customer service conversations, and telephone leads within the second response window. The intent submission stage records an intent-to-test-drive response when the device connects to test drive appointments, in-store check-ins, and test drive completion records within the third response window. The in-store test drive stage records a test drive-to-transaction response when the device connects to deposit, contract, invoicing, and delivery records within the fourth response window.

[0029] Furthermore, after the vehicle popularity version is generated, version verification is performed to verify the consistency between the vehicle object number and the evidence chain number, the consistency between the historical record time and the historical observation time range, the consistency between the late return record and the supplementary record period, the isolation status between the strong exposure non-conversion record and the vehicle's real demand evidence chain, the isolation status between the negative event attention record and the positive demand evidence, and the limitation status of the price chaos status and the inventory freeze status on the version output range.

[0030] When verification is successful, the status is registered as effective; when verification fails, the status is registered as failed and the reason for failure is written.

[0031] Compared with existing technologies, the present invention provides a method for generating vehicle heat maps based on historical data, which has the following beneficial effects:

[0032] 1. This invention establishes a vehicle popularity generation session, first defining the vehicle object number, historical observation time range, future demand observation cycle, region number, channel number, data cutoff time, and popularity configuration version number. Then, it aggregates vehicle behavior, conversion, price, inventory, exposure, and event history records according to the stages of front-end contact, interest retention, intention submission, in-store test drive, transaction formation, transaction withdrawal, and abnormal attention. It further registers the exposure impact status, event noise status, price status, and inventory status. Subsequently, it connects natural behavior records, back-end conversion records, and late return records according to the stage response window corresponding to the future demand observation cycle. By isolating records of high-exposure but no-conversion, negative event attention, and abnormal source records, the vehicle popularity version is no longer directly formed by the superposition of front-end data such as page views, clicks, and search volume. Instead, it is generated based on a connectable chain of evidence of real vehicle demand. In cases where the sampling time of multi-source historical data is inconsistent, conversion feedback is delayed, platform exposure self-excitation, and short-term event noise is superimposed, it can distinguish between short-term noise-type popularity, exposure-amplified popularity, negative attention-type popularity, and convertible sustainable popularity. This avoids misjudging advertising traffic, short-term attention, and negative public opinion as real vehicle demand, and reduces the underestimation of sustainable demand caused by delays in lead generation, test drives, and transaction feedback.

[0033] 2. This invention also registers vehicle popularity levels, popularity directions, popularity credibility status, popularity validity period, and version output range through vehicle popularity version registration. This gives vehicle popularity results source and usage boundaries. Popularity corresponding to records of strong exposure without conversion can be limited to the operational observation range, popularity corresponding to records of negative event attention can be limited to the risk observation range, vehicles with tight inventory and continuous backend conversion can enter the inventory allocation and sales follow-up range, and vehicles with chaotic pricing, frozen inventory, and missing key fields can enter the pending verification range. Through this versioned and status-based popularity output method, vehicle recommendation sorting, regional deployment, inventory allocation, and sales follow-up can read popularity results that match the intensity of actual demand, reducing the problems of exposure loop amplification, event noise mixing, and invalid popularity occupying business resources. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of a method for generating vehicle heat maps based on historical data according to the present invention. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] Example: Figure 1 A method for generating vehicle heat maps based on historical data is presented, including:

[0037] S1. Establish a vehicle popularity generation session, register the vehicle object number, historical observation time range, future demand observation period, region number, channel number, data cutoff time, and popularity configuration version number, and read vehicle behavior, conversion, price, inventory, exposure, and event history records, specifically:

[0038] Before generating vehicle heatmaps, a vehicle heatmap generation session must be established. Each vehicle heatmap generation session is formed in units of a heatmap generation task. When a session is established, the heatmap configuration version number is read, and the vehicle object scope, historical observation time range, future demand observation cycle, region number scope, channel number scope, data cutoff time, and readable historical record types are registered according to the heatmap configuration version number.

[0039] The vehicle object number is used to define the specific object for which vehicle popularity is generated. When the vehicle object is a single vehicle for sale, the vehicle object number is formed by concatenating the vehicle's unique listing number, the vehicle identification number (VIN) anonymized number, the sales store number, and the listing channel number in a fixed order. When the vehicle object is a combination of vehicle models and configurations, the vehicle object number is formed by concatenating the brand number, model series number, model year number, configuration number, energy type number, region number, and channel number in a fixed order.

[0040] The above-mentioned number fields are all read from vehicle listing records, vehicle model configuration records, store records, and channel records. When a field is missing, it is not directly merged into other vehicle objects. Instead, a missing placeholder is written in the corresponding position, and the missing field status of the object is registered in the vehicle popularity generation session.

[0041] The historical observation period is defined by both the start and end times. The start time is based on the observation period registered in the popularity configuration version, and the end time is the data cutoff time. If the actual listing time of a newly launched vehicle is later than the start time of the historical observation period, the actual listing time is used as the starting point for historical reading, and the short historical status of the new listing is recorded. Vehicles that have been removed from shelves, discontinued, or have frozen inventory and have exceeded the registration period of the popularity configuration version are included in the inventory review range and are not included in the regular positive popularity generation range.

[0042] The future demand observation period is used to define the applicable business period for the vehicle popularity version. The popularity configuration version registers one of the following periods: the next 7 days, the next 14 days, or the next 30 days. If no period is registered, the next 7 days are used and written to the default period state. The region code is read from the administrative division code, the store service area code, and the platform operation area code, prioritizing the regions corresponding to consultation stores, test drive stores, and transaction stores. If only browsing, clicking, and search records exist, the user access area is used; if the user access area is missing, the vehicle listing store area is used and registered as pending region verification.

[0043] Channel IDs are retrieved from natural search channels, platform recommendation channels, advertising channels, event entry channels, store private domain channels, third-party lead channels, and manual entry channels. If a channel field is missing, the channel is registered as pending verification. After the vehicle popularity generation session is established, vehicle behavior history, conversion history, price history, inventory history, exposure history, and event history are retrieved according to vehicle object ID, historical observation time range, region ID, channel ID, and data cutoff time.

[0044] Vehicle behavior history records include browsing, clicks, searches, favorites, model comparisons, configuration comparisons, price subscriptions, comments, and sharing records; conversion history records include inquiry leads, quote requests, customer service conversations, phone leads, test drive appointments, in-store check-ins, test drive completions, deposits, contracts, invoicing, delivery, appointment cancellations, order refunds, and order closure records; price history records include official guide price, platform display price, store price, discount amount, financial subsidies, trade-in subsidies, price version number, and price effective time; inventory history records include available quantity, quantity in transit, locked order quantity, sold but not delivered quantity, delisted quantity, inventory version number, and inventory update time; exposure history records include exposure batch number, entry type, page location, exposure start time, exposure end time, activity number, and exposed vehicle object number; event history records include event type, event occurrence time, event release channel, affected vehicle objects, event direction confirmation field, and event association records.

[0045] After reading, a raw historical record pool is formed. Each record in the raw historical record pool is registered with the original record number, data source, behavior occurrence time, data reception time, vehicle object number, region number, channel number, and data batch number. Records with both behavior occurrence time and data reception time missing are placed in the invalid isolation area; records with missing behavior occurrence time but existing data reception time are placed in the pending verification record area; records with data reception time later than the data deadline but within the supplementary recording period are placed in the late candidate area, providing a unified input for the historical record organization and phase collection in S2.

[0046] S2. Historical records are organized based on vehicle object number, behavior occurrence time, data reception time, channel number, exposure batch number, price version number, and inventory version number, and categorized into stages: front-end contact, interest retention, intention submission, in-store test drive, transaction formation, transaction withdrawal, and abnormal attention. Specifically:

[0047] Based on the original historical record pool formed by S1, the historical records are categorized and aggregated by stage. The categorization is based on the vehicle object number, the time of the behavior, the time of data reception, the channel number, the exposure batch number, the price version number, and the inventory version number as input fields, and a unified historical record is formed for each historical record.

[0048] The unified historical record registration includes vehicle object number, user anonymization number, device anonymization number, behavior type, behavior occurrence time, data reception time, channel number, entry type, page location, exposure batch number, activity number, price version number, inventory version number, region number, original record number, and record validity status.

[0049] When different data sources use different names for the same behavior, they are collectively grouped by the original behavior name, entry type, business result field, and subsequent business records. For example, inquiries, telephone consultations, online customer service, quotation requests, and lead generation records are all grouped into consultation leads when the business result field shows submitted contact information; test drive appointments, in-store appointments, and test drive registrations are all grouped into test drive appointments when the store number and appointment time exist; deposit payments, intention deposit payments, and order lock records are all grouped into deposits when the payment serial number exists; contract signing, invoicing, and delivery registration are linked together to form transaction records according to the business serial number.

[0050] When at least two of the original behavior name, entry type, business result field, and subsequent business record point to the same behavior type, the unified historical record is registered as that behavior type; when only one points to that behavior type, it is registered as behavior type pending verification; when different fields point to mutually exclusive behavior types, it is registered as behavior type conflict state, and it cannot subsequently enter the vehicle's true demand evidence chain.

[0051] When both the behavior occurrence time and the data reception time exist, the behavior occurrence time is used as the stage aggregation time; when the behavior occurrence time is missing but the data reception time exists, the data reception time is used to enter the pending record area, and the behavior time is recorded as missing; when the interval between the data reception time and the behavior occurrence time exceeds the late deadline registered in the heat configuration version, it is recorded as receiving overdue.

[0052] The data is aggregated in four stages: initial contact, interest retention, intention submission, in-store test drive, transaction formation, transaction cancellation, and unusual activity. The initial contact stage receives records of display, exposure, browsing, clicks, and search results; the interest retention stage receives records of favorites, model comparisons, configuration comparisons, and price subscriptions; the intention submission stage receives records of inquiry leads, price requests, customer service conversations, and phone leads; the in-store test drive stage receives records of test drive appointments, in-store check-ins, and test drive completion; the transaction formation stage receives records of deposits, contracts, invoicing, and delivery; the transaction cancellation stage receives records of appointment cancellations, order cancellations, refunds, and order closures; and the unusual activity stage receives records of complaints, recall inquiries, negative reviews, unusual sharing, and public opinion monitoring.

[0053] After the phased collection is completed, duplicate records of the same vehicle object number, the same user de-identified number, the same device de-identified number, and the same behavior type are merged and registered.

[0054] The initial contact phase uses a 30-minute repeat window. If the same user views and clicks on the same vehicle multiple times within 30 minutes, the first record is considered a valid contact record, and subsequent records are registered as repeat contact records. The interest retention phase uses a 24-hour repeat window. If the same user repeatedly saves and unsaves items within 24 hours, the interest retention status is recorded based on the last valid status, along with the number of status changes. The intention submission phase uses a 7-day repeat window. If the same user, contact information (anonymized ID), store number, and vehicle ID are used to submit multiple leads, the first valid lead is considered an intention submission, and subsequent leads are registered as repeat leads. The in-store test drive and transaction formation phases use the business transaction number. If the business transaction numbers are the same, the latest business status is used. If the business transaction numbers are different and the store numbers are different, or the time interval exceeds seven days, they are registered as independent conversion records.

[0055] When the user's de-identified ID is missing, the device's de-identified ID and the contact information's de-identified ID are used for duplicate identification; when all three are missing, the record can only enter the low-confidence candidate area.

[0056] After the above sorting, each available historical record has a unified vehicle object number, unified stage affiliation, unified time caliber, unified channel caliber, and unified record validity status, providing standardized input for S3 registration exposure impact status, event noise status, price status, and inventory status.

[0057] S3. Based on the exposure batch, entry type, page location, activity number, event type, price version, and inventory version, register the exposure impact status, event noise status, price status, and inventory status for each stage of historical records, specifically:

[0058] Based on the unified historical record and stage aggregation results generated by S2, the exposure impact status, event noise status, price status and inventory status are registered for each stage of historical record.

[0059] The exposure impact status is determined by reading the exposure batch number, entry type, page location, activity number, exposure start time, exposure end time, channel number, front-end contact stage record, and back-end conversion stage record. When the page location is an ad slot, homepage focus slot, first screen slot in the news feed, splash screen slot, or recommended slot, the corresponding front-end contact stage record is registered as a strong exposure contact record. When the activity number corresponds to a lottery, red envelope, points, task reward, or new user acquisition promotion, the corresponding front-end contact stage record is registered as an activity incentive contact record. When the entry type is natural search, natural entry from the vehicle model list, or natural entry from the store details page, and there is no exposure batch number or activity number, the corresponding record is registered as a natural behavior record.

[0060] When the exposure start time exists but the exposure end time is missing, the temporary exposure observation range is the calendar day after the exposure start time, and the exposure end point is recorded as pending. When the exposure end time exists but the exposure start time is missing, the temporary exposure observation range is the calendar day before the exposure end time, and the exposure start point is recorded as missing. When the exposure batch number is missing but the entry type and page location both point to the ad placement, the suspected exposure impact status is recorded.

[0061] The event noise status reads the event type, event occurrence time, event release channel, affected vehicle objects, event direction confirmation field, abnormal attention stage record, and transaction withdrawal stage record.

[0062] When the event type is quality complaint, recall, negative evaluation, delivery delay, or controversial news, the event direction is registered as a negative concern event; when the event type is official price reduction, financial subsidy, trade-in subsidy, or limited-time offer, the event direction is registered as a transaction stimulus event; when the event type is competitor product launch, competitor product price reduction, or competitor product recall, the event direction is registered as a competitor product disruption event; when the event type is inventory shortage, extended delivery cycle, or shortage of popular configurations, the event direction is registered as a supply restriction event.

[0063] If the event direction confirmation field exists, it will be used first; if the event direction confirmation field is missing, the event type, event publishing channel and related behavior records will be read together to register the event direction. If it still cannot be assigned, the event direction will be registered as pending verification.

[0064] When a negative attention event occurs, and searches, comments, shares, and views increase, while complaints, recall inquiries, cancellations, refunds, and appointment cancellations also occur, the corresponding records are registered as negative attention event records. When a transaction stimulus event occurs, and browsing, searches, inquiries, and test drives occur sequentially, the corresponding records are registered as transaction stimulus response records. When a competitor disturbance event occurs, and the same user makes comparison records between the target vehicle and competitor vehicles, competitor comparison records are registered.

[0065] The price status can be read as follows: price version number, official guide price, platform display price, store price, discount amount, financial subsidy, trade-in subsidy, price effective time, and price source.

[0066] When the platform display price, store price, and discount amount are consistent in direction in two consecutive price versions and the change range does not reach the registration threshold of the popularity configuration version, the registered price is in a stable state; when the platform display price decreases, the store price decreases, the discount amount increases, and the subsidy field takes effect, the registered price is in a downward state; when the platform display price increases, the store price increases, and the discount amount decreases, the registered price is in a rebound state; when there is a directional conflict between the official guide price, platform display price, and store price, the registered price is in a chaotic state; when key price fields are missing, the registered price is in a missing state.

[0067] The inventory status is read as follows: inventory version number, available quantity, quantity in transit, locked quantity, quantity sold but not delivered, quantity removed from shelves, inventory update time, and store availability flag.

[0068] When the available quantity reaches the registration threshold for the popularity configuration version and the number of locked orders does not reach the threshold, register the inventory as sufficient; when the available quantity is lower than the threshold, the number of locked orders increases, and there are sold but undelivered items, register the inventory as tight; when new available quantities are added to the inventory version, register the inventory as released; when there are available quantities of vehicles but any of the following states exist: frozen at the store, pending procedures, unsaleable, or delisted, register the inventory as frozen; when the inventory version number is missing, register the inventory as missing.

[0069] After the above four types of status registration are completed, each unified historical record carries source attribute, event attribute, price attribute and inventory attribute. S4 connects natural behavior records, backend conversion records and late return records based on this, and isolates strong exposure non-converted records, negative event attention records and source abnormal records.

[0070] S4. Based on the phase response window corresponding to the future demand observation cycle, connect the natural behavior records, backend conversion records, and late return records of the same vehicle object, and isolate the strong exposure non-conversion records, negative event attention records, and abnormal source records, specifically:

[0071] Based on the phase response window corresponding to the future demand observation cycle, the system connects the natural behavior records, backend conversion records, and late return records of the same vehicle object, while isolating records that do not enter the vehicle's actual demand evidence chain. The phase response window is registered by the popularity configuration version and corresponds to the future demand observation cycle.

[0072] When the future demand observation period is seven days, a 24-hour response window is used from the initial contact stage to the interest retention stage, a three-day response window is used from the interest retention stage to the intention submission stage, a seven-day response window is used from the intention submission stage to the in-store test drive stage, and a fourteen-day response window is used from the in-store test drive stage to the transaction formation stage. When the future demand observation period is fourteen days and thirty days, the response window for each stage is read according to the business cycle registered in the popularity configuration version.

[0073] The stage connection takes the vehicle object number, user de-identification number, device de-identification number, contact information de-identification number, region number, channel number, and behavior occurrence time as input.

[0074] First, read the natural behavior records under the same vehicle object number. Natural behavior records refer to records with channel number as natural search, natural entry from vehicle list, natural entry from store details, and no strong exposure contact state, activity incentive contact state, or abnormal source state.

[0075] Natural behavior records, arranged chronologically, form a user behavior sequence. Within this sequence, the initial contact stage is recorded when a user connects to favorites, car model comparisons, or price subscriptions within the first response window, registering a valid contact response. The interest retention stage is recorded when a user connects to inquiry leads, quote requests, customer service conversations, or phone lead records within the second response window, registering an interest-to-intent response. The intent submission stage is recorded when a user connects to test drive appointments, in-store check-ins, or test drive completion records within the third response window, registering an intent-to-test drive response. The in-store test drive stage is recorded when a user connects to deposit, contract, invoicing, or delivery records within the fourth response window, registering a test drive-to-transaction response.

[0076] The connection of backend conversion records takes the business serial number as the priority. When the business serial number is missing, the user's anonymized ID, contact information anonymized ID, store ID, vehicle object ID, and behavior occurrence time are read together to form a connection. When key fields are insufficient, the vehicle-level conversion is registered as pending verification and is not entered into the user-level connection chain.

[0077] Late return records come from the late candidate area in S1 and return data after the popular version takes effect. Late return records have the behavior occurrence time, data reception time, business serial number, and original record number. When the behavior occurrence time falls within the historical observation time range, the data reception time falls within the supplementary recording period, and the business serial number can be linked to existing consultation, test drive, and transaction records, it is registered as a late available record. When a late transaction record can be linked to at least one of the following: test drive appointment, store check-in, test drive completion, deposit, and contract, it is registered as a transaction late confirmation record. When a late record can only be linked to the vehicle object number and cannot be linked to the user's anonymized number, contact information anonymized number, and business serial number, it is registered as a vehicle-level late pending verification record.

[0078] Isolation rules are executed simultaneously during the connection process. Strong exposure without conversion records refer to strong exposure contact records that, within the corresponding stage response window, are not connected to the backend response records in the interest retention stage, intention submission stage, in-store test drive stage, and transaction formation stage. These records are retained in the exposure impact evidence area and do not enter the vehicle's actual demand evidence chain.

[0079] Negative event attention records refer to search, comment, sharing, and browsing records generated after a negative attention event occurs, and records of complaints, recall inquiries, unsubscriptions, refunds, and cancellations of appointments appear within the same time frame. These records are retained in the abnormal attention evidence area and are not used as positive demand evidence.

[0080] Source anomaly records refer to concentrated access within a short period of time from the same device with an anonymized ID, the same IP address, and the same activity entry point, without subsequent records of collection, inquiry, test drive, or transaction responses. These records enter the source anomaly evidence area. Isolated records still retain the original record number, vehicle object number, time of occurrence, and reason for isolation, and are subsequently used for popularity credibility registration and version verification, but do not enter the vehicle's true demand evidence chain. After phased connection and isolation, a vehicle's true demand evidence chain is formed.

[0081] The evidence chain for real vehicle demand is registered according to the contact evidence segment, interest evidence segment, intention evidence segment, test drive evidence segment, transaction evidence segment, withdrawal evidence segment, abnormal attention evidence segment, and supply and demand status evidence segment. Each evidence segment records the source, time range, stage response status, and abnormal marker, providing direct evidence for S5 to generate a vehicle popularity version.

[0082] S5. Based on the connected evidence chain of actual vehicle demand and its corresponding status, generate a vehicle popularity version, registering the popularity level, popularity direction, popularity credibility status, popularity validity period, and version output range, specifically:

[0083] Based on the evidence chain of real vehicle demand formed by S4 and its corresponding exposure impact status, event noise status, price status, inventory status, and late return records, a vehicle popularity version is generated.

[0084] Vehicle popularity versions are numbered using the vehicle object number, region number, channel number, data cutoff time, future demand observation period, and version number. When generating a version, the system first reads the stage response status in the evidence chain of actual vehicle demand, then reads the exposure impact status, event noise status, price status, and inventory status registered in S3, and finally reads the late availability records, late transaction confirmation records, vehicle-level late pending verification records, strong exposure non-conversion records, negative event attention records, and source anomaly records registered in S4. Popularity levels are formed according to the level table registered in the popularity configuration version.

[0085] When the evidence chain of genuine vehicle demand includes effective contact responses, interest-to-intent responses, intent-to-test-drive responses, and test-drive-to-transaction responses, and the withdrawal evidence segment does not reach the registration threshold for the popularity configuration version, the popularity level is registered as high-level. When the evidence chain of genuine vehicle demand has few front-end contact evidence segments, but intent evidence segments, test-drive evidence segments, and transaction evidence segments exist consecutively, and the inventory status is a tight inventory state, the popularity level should not be reduced to the lowest level based on low browsing criteria. When the evidence chain of genuine vehicle demand is mainly formed by strong exposure contact records, and back-end conversion records are missing, the popularity level is limited by strong exposure without conversion records. When the evidence chain of genuine vehicle demand is mainly formed by negative event attention records, the popularity level is not considered as positive demand popularity output.

[0086] Popularity is registered according to the stage response status and return status. When natural behavior records, intention submission records, in-store test drive records, and transaction formation records appear consecutively along the stage response window, the popularity direction is registered as rising; when the front-end contact stage fluctuates but the intention submission stage and in-store test drive stage are still within the response window, the popularity direction is registered as remaining; when transaction withdrawal stage records, refund records, inventory freeze status, and price chaos status persist, the popularity direction is registered as falling; when negative event attention records dominate, the popularity direction is registered as abnormal attention; when the absence of key fields causes the stage connection to be unable to be completed, the popularity direction is registered as pending verification.

[0087] The credibility status of popularity is registered according to the source of evidence. When the evidence chain of real vehicle demand is formed by natural behavior records and backend conversion records, and the consistency of object, time, source, and status are all verified, the credibility status of popularity is registered as confirmed. When strong exposure without conversion records exist and affect the front-end contact evidence segment, it is registered as affected by exposure. When negative event attention records exist and affect abnormal attention evidence segment, it is registered as affected by event. When price decline, price chaos, and price missing status affect behavioral response, it is registered as affected by price. When inventory shortage, inventory freeze, and inventory missing status affect the output range, it is registered as affected by inventory. When new listings with short history, a high proportion of missing user anonymized numbers, and vehicle-level late arrival records exist, it is registered as low sample or pending verification.

[0088] The validity period of the popularity index is determined by the data cutoff time and the future demand observation period. The first day after the data cutoff time is the start of the validity period, and the end date of the future demand observation period is the end date of the validity period.

[0089] For newly listed products with a short history, the validity period shall not exceed seven days; for products with frozen inventory, the validity period shall be registered as unsaleable pending verification; for products with a popularity and credibility status pending verification, the validity period shall only be used for review and shall not be included in high-level business output. The version output scope shall be registered according to the popularity and credibility status and popularity direction.

[0090] Confirmed rising and holding statuses can be output to vehicle recommendation ranking, sales follow-up, and regional demand dashboards; supply-constrained heat statuses can be output to inventory allocation and store replenishment reminders; exposure-affected statuses can be output to exposure assessment and operational observation; abnormal attention statuses can be output to risk observation and customer communication; pending verification statuses are only output to data verification.

[0091] After the vehicle popularity version is generated, version verification is performed to verify whether the vehicle object number is consistent with the evidence chain number, whether the historical record time is within the historical observation time range, whether the late return record is within the supplementary recording period, whether the strong exposure without conversion record is excluded from the vehicle real demand evidence chain, whether the negative event attention record is not used as positive demand evidence, and whether the price chaos state and inventory freeze state limit the corresponding output range.

[0092] When the verification is successful, the vehicle's heat version is registered as effective; when the verification fails, it is registered as a verification failure and the reason for the failure is written.

[0093] If a previous valid version exists and is still valid, the previous valid version will be retained; if the first version is generated and verification fails, it will be registered as having no valid popularity pending verification.

[0094] Once a vehicle popularity version takes effect, subsequent late transactions, order cancellations, price corrections, inventory corrections, and event direction corrections will be entered into the popularity backflow record and reprocessed in the next vehicle popularity generation session according to S1 to S5 to avoid overwriting the already effective vehicle popularity version without records.

[0095] In one specific embodiment, the historical observation period for the popularity configuration version registration is the past ninety days, the future demand observation period is seven days, and the data cutoff time is 24:00 on the same day.

[0096] The response window from the initial contact stage to the interest retention stage is 24 hours; the response window from the interest retention stage to the intention submission stage is 3 days; the response window from the intention submission stage to the in-store test drive stage is 7 days; the response window from the in-store test drive stage to the transaction completion stage is 14 days; and the deadline for supplementing late return records is 3 days.

[0097] When the same user's de-identified ID and the same device's de-identified ID generate multiple browsing and clicking records for the same vehicle within 30 minutes, the first record is retained as a valid contact record; when the same user repeatedly favorites and unfavorites within 24 hours, the interest retention status is registered based on the last valid status; when the same contact information's de-identified ID repeatedly submits inquiry leads for the same vehicle and the same store number within 7 days, the first valid lead is retained.

[0098] Within the same exposure batch, if the browsing and click records of the vehicle object within the exposure observation range reach more than twice the natural behavior benchmark, but no collection, model comparison, or price subscription records are generated within 24 hours, and no inquiry leads, quotation requests, or customer service conversation records are generated within three days, the exposure batch is registered as a strong exposure without conversion status.

[0099] If, within three days of a recall, quality complaint, or negative public opinion event, the search, comment, sharing, or browsing history for the same vehicle increases, and records of cancellations, refunds, appointment cancellations, complaint inquiries, or recall inquiries appear, and no inquiry leads, test drive appointments, or deposit records are generated within seven days, the corresponding record will be registered as a negative event attention record.

[0100] When the number of available vehicles under the same area code is less than three, and the number of locked orders reaches more than 50% of the available quantity, the inventory status is registered as "short inventory". When the platform display price, store price and discount amount all point to a lower transaction threshold in the same price version, the price status is registered as "price decline". When the official guide price, platform display price and store price show inconsistent upward and downward directions, the price status is registered as "price chaos".

[0101] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0102] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented in software, the above embodiments can be implemented in whole or in part by a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions of the embodiments of this application are implemented in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted wirelessly or wiredly from one website, computer, server, or data center to another website, computer, server, or data center. Wired methods include optical fiber, twisted pair, coaxial cable, etc. Wireless methods include infrared, microwave, etc. Available media include any available media that can be accessed by a computer or data storage devices such as servers and data centers that contain one or more sets of available media. Available media can be magnetic media (floppy disks, hard disks, magnetic tapes), optical media (DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0103] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for generating vehicle heat maps based on historical data, characterized in that, include: S1. Establish a vehicle popularity generation session, register the vehicle object number, historical observation time range, future demand observation cycle, region number, channel number, data cutoff time and popularity configuration version number, and read vehicle behavior, conversion, price, inventory, exposure and event history. S2. Organize historical records based on vehicle object number, behavior occurrence time, data reception time, channel number, exposure batch number, price version number, and inventory version number, and categorize them according to the stages of front-end contact, interest retention, intention submission, in-store test drive, transaction formation, transaction withdrawal, and abnormal attention. S3. Based on the exposure batch, entry type, page location, activity number, event type, price version, and inventory version, register the exposure impact status, event noise status, price status, and inventory status for the historical records of each stage. S4. According to the phase response window corresponding to the future demand observation cycle, connect the natural behavior records, backend conversion records and late return records of the same vehicle object, and isolate the strong exposure non-conversion records, negative event attention records and abnormal source records. S5. Generate a vehicle popularity version based on the evidence chain of real vehicle demand after connection and its corresponding status, and register the popularity level, popularity direction, popularity credibility status, popularity validity period and version output range.

2. The method of claim 1, wherein, The vehicle object number is formed by combining the unique listing number, the de-identified vehicle identification number, the sales store number, and the listing channel number in a fixed order under the single vehicle for sale category; and by combining the brand number, model number, model year number, configuration number, energy type number, region number, and channel number in a fixed order under the vehicle configuration combination category. When a field is missing, write a missing placeholder at the corresponding concatenation position and register the missing field status of the object.

3. The method of claim 1, wherein, When organizing historical records, inquiries, telephone consultations, online customer service, quotation requests and contact information are grouped into consultation leads; test drive appointments, in-store appointments and test drive registrations are grouped into test drive appointments; and deposit payments, intention deposit payments and order lock records are grouped into down payments. Contract signings, invoicing and delivery registrations are grouped into transaction formation records. The aggregation criteria are that at least two of the original behavior name, entry type, business result field, and subsequent business records point to the same behavior type.

4. The method of claim 1, wherein, After phased aggregation, duplicate merging is performed on historical records with the same vehicle object number, the same user de-identified number, the same device de-identified number, and the same behavior type. The initial contact phase uses a 30-minute repeat window and retains the first valid contact record; the interest retention phase uses a 24-hour repeat window and retains the last valid status; and the intention submission phase uses a 7-day repeat window and retains the first valid inquiry lead.

5. The method of claim 1, wherein, The impact status of exposure is registered based on the exposure batch number, entry type, page location, activity number, exposure start time, and exposure end time; When the page position is an advertisement position, a homepage focus position, a first screen position in the information feed, a splash screen position, or a recommendation position, the corresponding front-end contact stage is recorded as a high-exposure contact record. When an activity number corresponds to a lottery, red envelope, points, task rewards, or new user acquisition promotion, the corresponding front-end contact stage record is registered as an activity incentive contact record.

6. The method of claim 1, wherein, Event noise status is registered based on the event type, event occurrence time, event release channel, affected vehicles, and event direction confirmation fields; Quality complaints, recalls, negative reviews, delivery delays, and controversial news are registered as negative attention events; official price reductions, financial subsidies, trade-in subsidies, and limited-time offers are registered as transaction stimulus events; competitor product launches, competitor price reductions, and competitor product recalls are registered as competitor product disturbance events; and tight inventory, extended delivery cycles, and shortages of popular configurations are registered as supply restriction events.

7. The method of claim 1, wherein, Price status is registered based on price version number, official guide price, platform display price, store price, discount amount, financial subsidy, trade-in subsidy, and price effective time; When the platform display price, store price, and discount amount are in the same direction in two consecutive price versions and the change does not reach the registration threshold of the popularity configuration version, the registered price is in a stable state. When the platform display price decreases, the store price decreases, the discount amount increases, and the subsidy field takes effect, the registered price is in a downward state. When there is a conflict in the direction of increase or decrease among the official guide price, platform display price, and store price, the registered price is in a chaotic state.

8. The method of claim 1, wherein, Inventory status is registered based on inventory version number, available quantity, quantity in transit, locked order quantity, sold but not delivered quantity, removed from shelves quantity, inventory update time, and store availability flag. When the available quantity reaches the registration threshold for the popularity configuration version and the number of locked orders does not reach the threshold, the inventory status is registered as sufficient. When the available quantity is lower than the threshold, the number of locked orders increases, and there are sold but undelivered quantities, the inventory status is registered as tight. When at least one of the following exists, the inventory status is registered as frozen: store frozen, procedures pending, unsaleable, or delisted.

9. The method of claim 1, wherein, The phase response window is set according to the future demand observation cycle; The front-end contact stage records a valid response when the device connects to favorites, model comparisons, and price subscriptions within the first response window. The interest retention stage records an interest-to-intent response when the device connects to inquiry leads, quote requests, customer service conversations, and telephone leads within the second response window. The intent submission stage records an intent-to-test-drive response when the device connects to test drive appointments, in-store check-ins, and test drive completion records within the third response window. The in-store test drive stage records a test drive-to-transaction response when the device connects to deposit, contract, invoicing, and delivery records within the fourth response window.

10. The method of claim 1, wherein, After the vehicle popularity version is generated, version verification is performed to verify the consistency between the vehicle object number and the evidence chain number, the consistency between the historical record time and the historical observation time range, the consistency between the late return record and the supplementary record deadline, the isolation status between the strong exposure non-conversion record and the vehicle's real demand evidence chain, the isolation status between the negative event attention record and the positive demand evidence, and the limitation status of the price chaos status and inventory freeze status on the version output range. When verification is successful, the status is registered as effective; when verification fails, the status is registered as failed and the reason for failure is written.

Images