Intelligent mobile application optimization method based on online car-hailing driver behavior analysis
Through intelligent mobile applications, real-time collection and analysis of online ride-hailing driver data, using machine learning to optimize interfaces and paths, and formulate personalized incentives, solving the problem of difficult to improve driver and passenger experience in traditional methods, and achieving the effect of personalized optimization and continuous improvement.
Patent Information
- Application Number
- CN202510327411.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-19
- Publication Date
- 2025-07-08
AI Technical Summary
Traditional online ride-hailing driver behavior analysis methods are difficult to comprehensively and accurately improve the experience of drivers and passengers, and the evaluation of driver service quality is difficult to convey in a timely manner. The incentive mechanism is not targeted, which affects operational efficiency and user experience.
Through intelligent mobile applications, driver behavior data, passenger feedback data and application performance data are collected and analyzed in real time, and machine learning and deep learning algorithms are used to build driver portraits, optimize application interfaces and path selection, formulate personalized incentive plans, and monitor and adjust optimization strategies in real time.
It has achieved personalized optimization based on driver habits and preferences, improve application performance and user experience, improve driver enthusiasm and operational efficiency, and continuously improve application adaptability.
Smart Images

Figure CN120278380A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of mobile application optimization, and more specifically, relates to an intelligent mobile application optimization method based on the behavior analysis of online car-hailing drivers. Background Art
[0002] With the rapid development of the mobile Internet, online car-hailing services have become an important way for people's daily travel. Online car-hailing platforms connect drivers and passengers through mobile applications to provide convenient travel services. However, with the intensification of market competition, improving user experience and operational efficiency has become an important task for major online car-hailing platforms.
[0003] Online car-hailing drivers are the core of platform operation, and their behaviors and service qualities directly affect passengers' satisfaction and the platform's reputation. Traditional optimization methods often rely on manual experience and limited data, making it difficult to comprehensively and accurately improve the experiences of drivers and passengers.
[0004] In recent years, with the development of big data and artificial intelligence technologies, data analysis based on user behaviors has become an important means to improve product and service qualities. By analyzing the behavior data of online car-hailing drivers, their operation habits, service qualities, and work efficiencies can be deeply understood, and scientific optimization strategies can be formulated to improve the overall service level.
[0005] However, the current behaviors of online car-hailing drivers in various links such as receiving orders, driving, waiting, and completing orders are complex and diverse, making it difficult for traditional methods to comprehensively capture and analyze. In addition, drivers often encounter problems such as traffic congestion and unreasonable route selection during driving, which affect passengers' experiences and operational efficiencies. There are differences in the service qualities of different drivers, and it is difficult for passengers' feedback and evaluations to be conveyed to the platform in a timely manner, resulting in difficulties in improving service quality. Existing incentive mechanisms lack pertinence and are difficult to fully mobilize drivers' enthusiasm and improve their work efficiencies. The interface design and function settings of the driver-side application do not fully consider drivers' operation habits and actual needs, affecting the use experience.
[0006] In view of this, the present invention is specifically proposed. Summary of the Invention
[0007] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide an intelligent mobile application optimization method based on the behavior analysis of online car-hailing drivers, solving the problems raised in the above background art.
[0008] To solve the above technical problem, the basic concept of the technical solution adopted by the present invention is:
[0009] An intelligent mobile application optimization method based on the behavior analysis of online car-hailing drivers, comprising the following steps:
[0010] Collect the behavior data, driver profile data, passenger feedback data, and application performance data of online car-hailing drivers in real time through a smart mobile application;
[0011] Store the collected data in a cloud database and perform data cleaning, denoising, and formatting to ensure the accuracy and availability of the data;
[0012] Analyze the driver's behavior patterns, route selections, work habits, and passenger satisfaction based on machine learning, data mining, and deep learning algorithms;
[0013] Formulate corresponding optimization strategies based on the data analysis results to improve the operation experience and service quality of online car-hailing drivers;
[0014] Integrate the optimization strategies into the smart mobile application and perform real-time monitoring and dynamic adjustment.
[0015] Optionally, the driver behavior data includes operation records such as receiving orders, driving, waiting, and completing orders; the driver profile data includes age, gender, driving experience, and vehicle type information; the passenger feedback data includes ratings, comments, and complaint feedback content; the application performance data includes technical indicators such as loading time, response speed, and number of crashes.
[0016] Optionally, integrate a data monitoring module into the smart mobile application to record the operation information of the driver such as receiving orders, driving, waiting, completing orders, and canceling orders in real time, and combine it with information such as GPS data, time stamps, and driving trajectories;
[0017] During the driver registration and usage process, collect their basic information, including age, gender, driving experience, and vehicle type, and at the same time combine historical order receiving records, working hours, and dynamic service area data to build an accurate driver profile;
[0018] Obtain feedback information such as ratings, comments, and complaints from passengers about the driver through the built-in evaluation system, analyze the satisfaction of passengers with the driver's service, and use text analysis technology to extract key opinions;
[0019] Monitor the key performance indicators of the smart mobile application, including loading time, response speed, number of crashes, and memory occupancy, and record the running performance of the application in different network environments and devices.
[0020] Optionally, the steps for analyzing the satisfaction of passengers with the driver's service and using text analysis technology to extract key opinions are:
[0021] Define the average rating S of the driver as: where s i is the rating given by the i-th passenger, and N is the total number of ratings. This rating is used to evaluate the overall service quality of the driver. If the rating is consistently lower than the threshold Smin , it can trigger the warning or training mechanism;
[0022] The review text T of the passengers needs to be segmented and sentiment analyzed. Define the review set as: T = {t1, t2,..., t M}, where t i represents the i-th review. Conduct sentiment analysis on the reviews, and use the sentiment polarity score P t to calculate: where p j is the sentiment score of the j-th review, M is the total number of reviews. If P t is lower than the set threshold, there may be problems with the driver's service and further optimization is required;
[0023] The complaint data can be represented as a binary set: C = {(c1, w1), (c2, w2),..., (c K , w K )}
[0024] where c k represents the k-th type of complaint, w k is the weight of this category, and calculate the comprehensive complaint severity C s , where f k is the occurrence times of the k-th type of complaint. If C s exceeds the set threshold C max , corresponding intervention measures need to be taken, such as warnings, training, or suspension of operation;
[0025] Through the TF-IDF method, extract high-frequency keywords and calculate the weight of a certain keyword w in all reviews. Among them, the word frequency represents the frequency of a certain word appearing in the review text: The inverse document frequency is used to measure the discrimination of this word in all evaluation data, and the calculation formula is: where N is the total number of all reviews, and n w is the number of reviews containing this word w;
[0026] Combined with the scoring data, sentiment analysis, and complaint severity, calculate the comprehensive satisfaction S final of the driver, S final = αS + βP t - γC s , where α, β, γ are adjustment weights to ensure the reasonable integration of scoring, sentiment analysis, and complaint information. If S final is lower than the set threshold, the driver needs to optimize the service.
[0027] Optionally, based on machine learning, data mining, and deep learning algorithms, the steps for analyzing the driver's behavior pattern are:
[0028] Classify the driving behavior patterns of drivers using the clustering analysis method. Assume that the driver behavior data set is D = {x1, x2,..., x N}, where each data point x i contains the operation records of the driver, and use the K-Means algorithm for clustering;
[0029] Among them, K is the preset number of clusters, c j is the cluster center, w ij is an indicator variable, which is 1 if x i belongs to cluster j, and 0 otherwise.
[0030] Optionally, the steps for the driver's route selection are as follows:
[0031] The driver's route selection is affected by factors such as historical driving trajectories, real-time road conditions, and passenger demands. Use the reinforcement learning method to optimize the route. Define the state s t as the current road condition, the action a t as the driver's route selection, and the reward function R t reflects the order completion rate and passenger satisfaction. Then the optimization goal is to maximize the expected return: Among them, γ is the discount factor, which is used to balance short-term and long-term benefits; use the Deep Q-Network (DQN) method to iteratively update the route selection strategy through Q-learning:
[0032] Finally, generate a personalized route recommendation system to help drivers choose the optimal driving route and improve the order acceptance efficiency.
[0033] Optionally, the steps for analyzing the driver's work habits are as follows:
[0034] Use time series analysis to predict the driver's work pattern. Define the daily order acceptance volume of the driver as the time series {x t}, and use the Long Short-Term Memory network for modeling: h t = σ(W h h t-1 + W x x t + b), where h t is the hidden state, W h , W x are the weight matrices, and b is the bias term.
[0035] Optionally, the steps for analyzing the driver's passenger satisfaction are as follows:
[0036] Combine passenger ratings, comments, and complaint data to build a sentiment analysis model and use BERT for text sentiment classification;
[0037] Assume that the passenger comment text is T = {t1, t2, ..., t M}, and the implicit semantics is calculated through the BERT encoding layer:
[0038] H = BERT(T) is then classified through Softmax: P(y|T) = softmax(WH+b), where W is the weight matrix, b is the bias term, and P(y|T) outputs the sentiment category of the comment;
[0039] Use the SHAP method to analyze the contribution of different features to driver scores: Among them, S(f) is the contribution of feature f to the score, F is the entire feature set, and V(S) is the contribution value of subset S.
[0040] Optionally, based on the data analysis results, the steps for formulating corresponding optimization strategies to improve the operating experience and service quality of online ride-hailing drivers include:
[0041] Based on the analysis results of the driver's behavior pattern, the interface layout and interaction design of the mobile application are optimized to improve the convenience and efficiency of operation. Assume that the driver's operation time is T o and operation success rate S o There is a negative correlation between them, so the optimization objective is:
[0042] Based on the path analysis results, the navigation function is improved to provide drivers with better route planning. Assuming that the driver chooses path p at time t i , the path weight function is defined as: C(p i )=αT i +βD i +γO i , where T i is the estimated travel time, D i is the driving distance, O i is the congestion coefficient, α, β, γ are weight parameters, and the optimization objective is: Utilize the analysis results of drivers’ working habits to develop personalized incentive plans to improve drivers’ motivation and order completion rate;
[0043] The driver's incentive benefit function can be defined as: Among them, A i is the driver's behavior indicator, w i is the weight parameter;
[0044] The driver's satisfaction score function is defined as: Among them, F j represents the jth feedback dimension, ω j As weight; Combine the driver's behavior pattern and work habits to optimize the message push mechanism;
[0045] Define the optimization objective of push messages: Among them, CTR(M) is the click-through rate, D(M) is the interference degree, λ is the balance parameter, and the collaborative filtering method is used to calculate the preference degree of drivers for different message types to optimize the push content and time: Among them, P(u,i) is the predicted preference value of user u for message i, μ is the global mean, b u , b i is the deviation between the user and the message, q i , p u is the feature vector;
[0046] Deploy the optimization strategy to the intelligent mobile application, and establish a real-time monitoring system, and set the key performance indicators: KPI = {O r , S d , C p , U e}, where O r is the order completion rate, S d is the driver satisfaction score, C p is the passenger complaint rate, U e is the application usage efficiency.
[0047] After adopting the above technical solutions, the present invention has the following beneficial effects compared with the prior art. Of course, any product implementing the present invention does not necessarily need to achieve all the advantages described below:
[0048] The intelligent mobile application optimization system based on user behavior analysis of the present invention can effectively perform personalized optimization according to the usage habits and preferences of users, improving the performance and user experience of the application. Through the real-time feedback and dynamic adjustment mechanism, the system can continuously improve and better meet the changing needs of users, providing strong technical support for the continuous development of mobile applications. The present invention proposes an intelligent mobile application optimization system based on the behavior analysis of online car-hailing drivers. The system collects and analyzes the behavior data of drivers, and through technologies such as artificial intelligence and machine learning, combined with driver portraits and passenger feedback, deeply analyzes the mobile application usage patterns of users, and provides data-driven optimization strategies. The system can automatically identify the operation habits, preference settings and usage scenarios of users, and then adjust the application interface, function layout and resource allocation to improve the user experience and the operation efficiency of the application.
[0049] The following further describes in detail the specific implementation manners of the present invention with reference to the accompanying drawings. Description of the Drawings
[0050] The accompanying drawings in the following description are only some embodiments. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. In the attached
[0051] In the figures:
[0052] Figure 1 It is a flowchart of an intelligent mobile application optimization method.
[0053] It should be noted that these drawings and the text description are not intended to limit the scope of the concept of the present invention in any way, but to illustrate the concept of the present invention to those skilled in the art by referring to specific embodiments. Specific embodiments
[0054] Now, the present invention will be further described in detail with reference to the accompanying drawings.
[0055] Please refer to Figure 1 As shown, in this embodiment, an intelligent mobile application optimization method based on the analysis of online car-hailing driver behavior is provided, including the following steps:
[0056] Real-time collect the behavior data, driver portrait data, passenger feedback data and application performance data of online car-hailing drivers through the intelligent mobile application;
[0057] Store the collected data in the cloud database, and perform data cleaning, denoising, and formatting processing to ensure the accuracy and availability of the data; at the same time, adopt data encryption and access control technologies to ensure the security and privacy protection of the data. In addition, standardize and extract features from the data to support subsequent data analysis and optimization strategy formulation.
[0058] Based on machine learning, data mining, and deep learning algorithms, analyze the driver's behavior patterns, route selection, work habits, and passenger satisfaction; behavior pattern analysis is used to identify the driver's operation habits and order-taking preferences; route optimization analysis combines real-time traffic data and historical driving records to optimize the driving route to reduce congestion time; work habit analysis evaluates the driver's working hours, order-taking frequency, and driving efficiency to formulate personalized order-taking suggestions; satisfaction analysis evaluates the driver's service quality based on passenger feedback data and identifies the key factors affecting passenger satisfaction.
[0059] Based on the data analysis results, formulate corresponding optimization strategies to improve the operation experience and service quality of online car-hailing drivers; the interface optimization strategy adjusts the application interface layout according to the driver's operation habits to improve the convenience of interaction; the route optimization strategy provides intelligent navigation suggestions based on the route analysis results to reduce driving time and fuel consumption; the incentive mechanism optimization strategy formulates personalized reward programs according to the driver's behavior patterns, such as peak period rewards, long-term service rewards, etc., to improve the driver's enthusiasm; the push strategy optimization plan combines the driver's usage habits to intelligently push important information to improve the acceptance rate and effectiveness of information.
[0060] Integrate the optimization strategies into the intelligent mobile application and conduct real-time monitoring and dynamic adjustment. The strategy deployment ensures that the optimization plan can be successfully applied to the mobile terminal; the real-time monitoring module continuously tracks the execution of the optimization strategy, collects feedback data from drivers and passengers, and adjusts abnormal situations; the effect evaluation module regularly evaluates the actual effect of the optimization plan based on key operation indicators (order acceptance rate, completion rate, driver satisfaction, passenger satisfaction, etc.), and continuously optimizes and improves according to the evaluation results to enhance the overall usage experience and operation efficiency.
[0061] In this embodiment, the driver behavior data includes operation records such as order acceptance, driving, waiting, and order completion; the driver portrait data includes information such as age, gender, driving experience, and vehicle type; the passenger feedback data includes scores, comments, and complaint feedback content; the application performance data includes technical indicators such as loading time, response speed, and number of crashes.
[0062] In this embodiment, integrate a data monitoring module in the intelligent mobile application to record the operation information of the driver's order acceptance, driving, waiting, order completion, order cancellation, etc. in real time, and combine information such as GPS data, timestamp, and driving trajectory to ensure the integrity and accuracy of the behavior data;
[0063] During the driver registration and usage process, collect their basic information, including age, gender, driving experience, vehicle type, etc., and at the same time combine dynamic data such as historical order acceptance records, working hours, and service areas to build an accurate driver portrait;
[0064] Through the built-in evaluation system, obtain feedback information such as the scores, comments, and complaints of passengers about the driver, analyze the satisfaction of passengers with the driver's service, and use text analysis technology to extract key opinions;
[0065] Monitor the key performance indicators of the intelligent mobile application, including loading time, response speed, number of crashes, memory occupancy, etc., and record the running performance of the application in different network environments and devices.
[0066] In this embodiment, the steps of analyzing the satisfaction of passengers with the driver's service and using text analysis technology to extract key opinions are:
[0067] Define the average rating S of the driver as: where s i is the rating given by the i-th passenger, N is the total number of ratings, and this rating is used to evaluate the overall service quality of the driver. If the rating is consistently lower than the threshold S min , a warning or training mechanism can be triggered.
[0068] The comment text T of the passenger needs to be segmented and sentiment analyzed. Define the comment set as: T = {t1, t2,..., t M}, where t i represents the i-th comment. Perform sentiment analysis on the comment and calculate using the sentiment polarity score P t : where p j is the sentiment score of the j-th comment (ranging from -1 to 1, with negative values representing negative emotions and positive values representing positive emotions), and M is the total number of comments. If P t is lower than the set threshold, there may be problems with the driver's service and further optimization is required.
[0069] The complaint data can be represented as a binary set: C = {(c1, w1), (c2, w2),..., (c K , w K )}
[0070] where c k represents the k-th type of complaint (such as bad attitude, route deviation, being late, etc.), and wkw_kwk is the weight of this category (set based on historical data and user feedback). Calculate the comprehensive complaint severity C s ,
[0071] where f k is the occurrence times of the k-th type of complaint. If C s exceeds the set threshold C max , corresponding intervention measures need to be taken, such as warning, training, or suspension of service.
[0072] Through the TF-IDF method, extract high-frequency keywords and calculate the weight of a certain keyword w in all comments. Among them, the term frequency represents the frequency of a word appearing in the comment text: The inverse document frequency is used to measure the distinctiveness of the word in all evaluation data, and the calculation formula is: where N is the total number of all comments, and n w is the number of comments containing the word w. Words with high TF-IDF values are the key opinions on the driver's service quality and can be used to further optimize the service.
[0073] Calculate the comprehensive satisfaction S of the driver by combining the scoring data, sentiment analysis, and complaint severity final , S final =αS + βP t -γC s , where α, β, and γ are adjustment weights to ensure the reasonable integration of scoring, sentiment analysis, and complaint information. If S final is lower than the set threshold, the driver needs to optimize the service, which may trigger training or intervention measures.
[0074] In this embodiment, the steps of analyzing the driver's behavior pattern based on machine learning, data mining, and deep learning algorithms are as follows:
[0075] Use the clustering analysis method to classify the driver's behavior pattern. Assume the driver behavior data set is D = {x1, x2,..., x N}, where each data point x i contains the driver's operation records (order receiving frequency, driving time, order completion rate, etc.), and the K-Means algorithm is used for clustering:
[0076] where K is the preset number of clusters, c j is the cluster center, and w ij is an indicator variable, which is 1 if x i belongs to cluster j, otherwise 0. Through clustering, different types of driver groups are identified, such as efficient drivers, low-active drivers, novice drivers, etc., and corresponding optimization strategies are formulated.
[0077] In this embodiment, the steps for the driver's route selection are as follows:
[0078] The driver's route selection is affected by factors such as historical driving trajectories, real-time road conditions, and passenger demands. The reinforcement learning method is used to optimize the route. Define the state s t as the current road condition, the action a t as the driver's route selection, and the reward function R t reflects the order completion rate and passenger satisfaction. Then the optimization goal is to maximize the expected return: where γ is the discount factor used to balance short-term and long-term benefits.
[0079] Use the Deep Q-Network (DQN) method to iteratively update the route selection strategy through Q-learning: Finally, a personalized route recommendation system is generated to help the driver select the optimal driving route and improve the order receiving efficiency.
[0080] In this embodiment, the steps of analyzing the driver's work habits are as follows:
[0081] Predict the driver's working pattern using time series analysis. Define the daily order-taking volume of the driver as the time series {x t}, and use a long short-term memory network for modeling: h t = σ(W h h t-1 + W x x t + b), where h t is the hidden state, W h , W x are the weight matrices, and b is the bias term.
[0082] The LSTM predicts the high-activity periods of the driver and provides personalized order-taking suggestions, such as reminding the driver to increase the online duration during peak hours, optimizing the push strategy, and improving the order-taking rate.
[0083] The steps for analyzing the passenger satisfaction of the driver in this embodiment are as follows:
[0084] Combine the passenger rating, review, and complaint data to build a sentiment analysis model, and use BERT for text sentiment classification;
[0085] Assume the passenger review text is T = {t1, t2,..., t M}, and calculate the hidden semantics through the BERT encoding layer:
[0086] H = BERT(T) Then, through Softmax classification: P(y|T) = softmax(WH + b), where W is the weight matrix, b is the bias term, and P(y|T) outputs the sentiment category (positive, neutral, negative) of the review.
[0087] When the proportion of negative reviews is too high, the system can automatically generate improvement suggestions to improve the driver's service quality;
[0088] Use the SHAP method to analyze the contribution degree of different features to the driver rating: where S(f) is the contribution degree of feature f to the rating, F is the set of all features, and V(S) is the contribution value of subset S.
[0089] The analysis results can be used to optimize the driver incentive mechanism. For example, provide training for drivers with a declining rating and offer additional rewards for drivers with a high rating.
[0090] The steps for formulating corresponding optimization strategies based on the data analysis results to improve the operation experience and service quality of online car-hailing drivers in this embodiment include:
[0091] Based on the analysis results of the driver's behavior pattern, optimize the interface layout and interaction design of the mobile application to improve the operation convenience and efficiency. Assume the operation time T o of the driver and the operation success rate So There is a negative correlation, and the optimization goal is:
[0092] The interface optimization measurement defines the interface optimization function U(x) as a function of the interaction time, where represents the interface complexity, and the optimization goal is to find the optimal x * such that: Using the Reinforcement Learning (RL) method, adjust the interface layout according to the driver's usage habits, and adopt Policy Gradient optimization: where π(a|s, θ) is the policy function, and R is the optimized operation efficiency benefit.
[0093] Based on the path analysis results, improve the navigation function and provide a better route plan for the driver. Assume that the driver selects path p at time t i , and the path weight function is defined as:
[0094] C(p i ) = αT i + βD i + γO i where T i is the estimated travel time, D i is the travel distance, O i is the congestion coefficient, and α, β, γ are weight parameters. The optimization goal is:
[0095] The optimization methods include:
[0096] Dynamically adjust the path based on the Dijkstra or A algorithm *
[0097] Combined with historical data and real-time road conditions, use LSTM to predict the traffic state within the next 5 - 10 minutes to improve the accuracy of path recommendations
[0098] Using the analysis results of the driver's work habits, formulate a personalized incentive plan to improve the driver's enthusiasm and order completion rate. The driver's incentive benefit function can be defined as: where A i is the driver's behavior index (such as order acceptance rate, positive review rate, driving mileage, etc.), and w i is the weight parameter.
[0099] Based on SHAP attribution analysis, determine the factors that have the greatest impact on performance and optimize the incentive measures:
[0100] For drivers who accept orders during peak hours, increase the dynamic bonus ratio
[0101] For drivers with high satisfaction, provide long-term service rewards
[0102] Combining LSTM to predict work habits, push targeted order - receiving suggestions, and improve the willingness to receive orders. Based on passenger satisfaction analysis, provide targeted optimization strategies for drivers with low ratings.
[0103] Define the driver's satisfaction score function as: Among them, F j represents the j - th feedback dimension (such as service attitude, driving smoothness, interior cleanliness, etc.), and ω j is the weight.
[0104] The optimization strategies include:
[0105] Push online training courses to drivers with low ratings to improve service awareness
[0106] Automatically identify negative - comment keywords, such as "bad attitude" and "taking a detour", and generate personalized improvement suggestions
[0107] Give extra exposure to drivers with high ratings, such as preferentially allocating high - quality orders.
[0108] Combine the driver's behavior patterns and work habits to optimize the message - pushing mechanism, reduce interference, and improve the accuracy of pushing.
[0109] Define the optimization goal of the pushed message: Among them, CTR(M) is the click - through rate, D(M) is the degree of interference, and λ is the balance parameter.
[0110] Adopt the collaborative filtering method to calculate the driver's preference for different message types, and optimize the pushing content and time: Among them, P(u, i) is the predicted preference value of user u for message i, μ is the global mean, b u , b i is the deviation of the user and the message, q i , p u are the feature vectors.
[0111] The optimization solutions include:
[0112] Reduce the pushing of low - relevance messages to avoid interfering with the driver's operations;
[0113] Push key notifications during the driver's high - activity period to improve the response rate;
[0114] Customize the pushing content for different types of drivers (newbies, seniors, efficient).
[0115] Deploy the optimization strategies to the intelligent mobile application and establish a real - time monitoring system to ensure the effectiveness of the optimization solutions.
[0116] Set key performance indicators: KPI = {Qr , S d , C p , U e}, where O r is the order completion rate, S d is the driver satisfaction score, C p is the passenger complaint rate, U e is the application usage efficiency.
[0117] The present invention is not limited to the above embodiments. Anyone should know that structural changes made under the inspiration of the present invention, as long as they have the same or similar technical solutions as the present invention, fall within the protection scope of the present invention. The technologies, shapes, and structures not described in detail in the present invention are all well-known technologies.
Claims
1. An intelligent mobile application optimization method based on the behavior analysis of online car-hailing drivers, characterized in that, It includes the following steps: Collect the behavior data, driver profile data, passenger feedback data, and application performance data of online car-hailing drivers in real time through a smart mobile application; Store the collected data in a cloud database and perform data cleaning, denoising, and formatting to ensure the accuracy and availability of the data; Analyze the driver's behavior patterns, route selection, work habits, and passenger satisfaction based on machine learning, data mining, and deep learning algorithms; Based on the data analysis results, formulate corresponding optimization strategies to improve the operation experience and service quality of online car-hailing drivers; Integrate the optimization strategies into the smart mobile application and perform real-time monitoring and dynamic adjustment.
2. The intelligent mobile application optimization method based on the behavior analysis of online car-hailing drivers according to claim 1, wherein Driver behavior data includes operation records such as receiving orders, driving, waiting, and completing orders; driver profile data includes age, gender, driving experience, and vehicle type information; passenger feedback data includes ratings, comments, and complaint feedback content; application performance data includes technical indicators such as loading time, response speed, and number of crashes.
3. An intelligent mobile application optimization method based on the behavior analysis of online car-hailing drivers according to claim 1, characterized in that, Integrate a data monitoring module into the smart mobile application to record the operation information of drivers such as receiving orders, driving, waiting, completing orders, and canceling orders in real time, and combine it with information such as GPS data, timestamp, and driving trajectory; During the driver registration and usage process, collect their basic information, including age, gender, driving experience, and vehicle type, and combine historical order receiving records, working hours, and dynamic service area data to build an accurate driver profile; Through the built-in evaluation system, obtain feedback information such as ratings, comments, and complaints from passengers about the driver, analyze the satisfaction of passengers with the driver's service, and use text analysis technology to extract key opinions; Monitor the key performance indicators of the smart mobile application, including loading time, response speed, number of crashes, and memory occupancy, and record the operation performance of the application in different network environments and devices.
4. An intelligent mobile application optimization method based on the behavior analysis of online car-hailing drivers according to claim 1, characterized in that, The steps for analyzing the satisfaction of passengers with the driver's service and using text analysis technology to extract key opinions are: Define the average rating S of the driver as: where s i is the rating given by the i-th passenger, N is the total number of ratings, and this rating is used to evaluate the overall service quality of the driver. If the rating is consistently lower than the threshold S min , a warning or training mechanism can be triggered; The review text T of the passengers needs to be segmented and sentiment analyzed. Define the review set as: T = {t1, t2,..., t M}, where t i represents the i-th review. Conduct sentiment analysis on the reviews and calculate using the sentiment polarity score P t : where p j is the sentiment score of the j-th review, M is the total number of reviews. If P i is lower than the set threshold, there may be problems with the driver's service and further optimization is required; Complaint data can be represented as a binary set: C = { (c1, w1), (c2, w2),..., (c K , w K )} Among them, c k represents the k-th type of complaint, and w k is the weight of this category. Calculate the comprehensive complaint severity C s , where f k is the occurrence times of the k-th type of complaint. If C s exceeds the set threshold C max , corresponding intervention measures need to be taken, such as warnings, training, or suspension of operation; By using the TF-IDF method, high-frequency keywords are extracted, and the weight of a certain keyword w in all comments is calculated. Among them, the term frequency represents the frequency of a word appearing in the comment text: The inverse document frequency is used to measure the discrimination of the word in all evaluation data, and the calculation formula is: Among them, N is the total number of all comments, n w is the number of comments containing the word w; Calculate the comprehensive satisfaction S of the driver by combining the scoring data, sentiment analysis, and complaint severity final , S final = αS + βP t - γC s , where α, β, and γ are adjustment weights to ensure the reasonable integration of scoring, sentiment analysis, and complaint information. If S final is lower than the set threshold, then the driver needs to optimize their service.
5. An intelligent mobile application optimization method based on the behavior analysis of online car-hailing drivers according to claim 1, characterized in that, The steps for analyzing the driver's behavior patterns based on machine learning, data mining, and deep learning algorithms are: Classify the driver's behavior patterns using the clustering analysis method. Assume that the driver behavior data set is D = {x1, x2,..., x N}, where each data point x i contains the driver's operation records, and the K-Means algorithm is used for clustering; where K is the preset number of clusters, c j is the cluster center, w ij is an indicator variable that is 1 if x i belongs to cluster j and 0 otherwise.
6. The intelligent mobile application optimization method based on the behavior analysis of online car-hailing drivers according to claim 1, wherein The steps for analyzing the driver's route selection are: The driver's route selection is affected by factors such as historical driving trajectories, real-time road conditions, and passenger demands. The reinforcement learning method is used to optimize the route. Define the state s t as the current road condition, and the action a t as the driver's route selection. The reward function R t reflects the order completion rate and passenger satisfaction. Then the optimization goal is to maximize the expected return: where γ is the discount factor used to balance short-term and long-term rewards. The deep Q-network (DQN) method is used to iteratively update the route selection strategy through Q-learning: Finally, a personalized route recommendation system is generated to help drivers select the optimal driving route and improve the order-taking efficiency.
7. An intelligent mobile application optimization method based on the behavior analysis of online car-hailing drivers according to claim 1, characterized in that, The steps for analyzing the driver's work habits are: Predict the driver's working pattern using time series analysis. Define the daily order reception volume of the driver as the time series {x t}, and use a long short-term memory network for modeling: h t = σ(W h h t-1 + W x x t + b), where h t is the hidden state, W h , W x are the weight matrices, and b is the bias term.
8. An intelligent mobile application optimization method based on the behavior analysis of online car-hailing drivers according to claim 1, characterized in that The steps for analyzing the driver's passenger satisfaction are: Combine passenger rating, comment, and complaint data to build a sentiment analysis model and use BERT for text sentiment classification; Assume that the passenger review text is \(T = \{t_1, t_2, \ldots, t\}\), M After the BERT encoding layer calculates the implicit semantics: \(H = BERT(T)\) Then classify through Softmax: \(P(y|T)=\text{softmax}(WH + b)\), where \(W\) is the weight matrix, \(b\) is the bias term, and \(P(y|T)\) outputs the sentiment category of the review; Using the SHAP method to analyze the contribution of different features to the driver's score: Among them, S(f) is the contribution of feature f to the score, F is the set of all features, and V(S) is the contribution value of subset S.
9. An intelligent mobile application optimization method based on the analysis of online car-hailing driver behavior according to claim 1, characterized in that, The steps for formulating corresponding optimization strategies based on the data analysis results to improve the operation experience and service quality of online car-hailing drivers include: Based on the analysis results of the driver's behavior patterns, optimize the interface layout and interaction design of the mobile application to improve the operation convenience and efficiency. Assume the driver's operation time T o and the operation success rate S o have a negative correlation. Then the optimization goal is: Based on the path analysis results, improve the navigation function to provide drivers with better route planning. Assume that the driver selects path p at time t i , the path weight function is defined as: C(p i ) = αT i + βD i + γO i , where T i is the estimated driving time, D i is the driving distance, O i is the congestion coefficient, and α, β, γ are weight parameters. The optimization objective is: Utilize the analysis results of drivers' work habits to formulate personalized incentive programs to improve drivers' enthusiasm and order completion rate; The incentive revenue function of the driver can be defined as: where A i is the driver's behavior index, and w i is the weight parameter; Define the satisfaction score function of the driver as follows: where F j represents the j-th feedback dimension, and ω j is the weight; Combine the driver's behavior patterns and work habits to optimize the message push mechanism; Define the optimization objectives for push messages: Among them, CTR(M) is the click-through rate, D(M) is the interference degree, λ is the balance parameter. Using the collaborative filtering method, calculate the preference degrees of drivers for different message types, and optimize the push content and time: Among them, P(u, i) is the predicted preference value of user u for message i, μ is the global mean, b u , b i is the deviation between the user and the message, q i , p u are feature vectors; Deploy the optimization strategy into the intelligent mobile application and establish a real-time monitoring system, and set the key performance indicators: KPI = {O r , S d , C p , U e}, where O r is the order completion rate, S d is the driver satisfaction score, C p is the passenger complaint rate, and U e is the application usage efficiency.