CN117236527A - Automobile part demand prediction method and system based on ensemble learning - Google Patents
Automobile part demand prediction method and system based on ensemble learning Download PDFInfo
- Publication number
- CN117236527A CN117236527A CN202311502410.1A CN202311502410A CN117236527A CN 117236527 A CN117236527 A CN 117236527A CN 202311502410 A CN202311502410 A CN 202311502410A CN 117236527 A CN117236527 A CN 117236527A
- Authority
- CN
- China
- Prior art keywords
- data
- emotion
- automobile part
- model
- demand
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 230000008451 emotion Effects 0.000 claims abstract description 111
- 238000012795 verification Methods 0.000 claims abstract description 46
- 238000012360 testing method Methods 0.000 claims abstract description 38
- 238000005065 mining Methods 0.000 claims abstract description 36
- 238000012549 training Methods 0.000 claims abstract description 33
- 238000012545 processing Methods 0.000 claims abstract description 26
- 230000006870 function Effects 0.000 claims abstract description 17
- 230000011218 segmentation Effects 0.000 claims abstract description 13
- 238000004140 cleaning Methods 0.000 claims description 22
- 238000004458 analytical method Methods 0.000 claims description 16
- 238000007418 data mining Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 10
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000010187 selection method Methods 0.000 claims description 9
- 230000004931 aggregating effect Effects 0.000 claims description 8
- 238000013480 data collection Methods 0.000 claims description 8
- 238000012423 maintenance Methods 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 5
- 238000002790 cross-validation Methods 0.000 claims description 5
- 238000012935 Averaging Methods 0.000 claims description 3
- 230000009193 crawling Effects 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 3
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 238000011835 investigation Methods 0.000 claims description 3
- 238000000513 principal component analysis Methods 0.000 claims description 3
- 238000007637 random forest analysis Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 238000013500 data storage Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an automobile part demand prediction method and system based on ensemble learning, wherein the method comprises the following steps: and collecting automobile part demand data, processing the data, mining emotion data, constructing an automobile part demand prediction model and optimizing the automobile part demand prediction model. The invention belongs to the technical field of data processing, and particularly relates to an automobile part demand prediction method and system based on ensemble learning, wherein the scheme adopts an emotion mining model, calculates an average emotion value in an observation period, calculates a function of the sales quantity of parts changing along with time, and evaluates the effectiveness of the emotion mining model; adopting an automobile part demand prediction model to perform integrated feature selection, calculating comprehensive weights of features, establishing an improved stacking frame, and adjusting parameters of a base learner by using grid search cross verification; and adopting data segmentation, and splicing a training set, a verification set and a test set of the base learner to the element learner to obtain a more stable prediction result.
Description
Technical Field
The invention belongs to the technical field of data processing, and particularly relates to an automobile part demand prediction method and system based on ensemble learning.
Background
The automobile part demand prediction is a method for predicting the future automobile part demand by collecting and analyzing historical sales data and market trend data, and aims to help enterprises reasonably arrange production plans, optimize inventory processing and reduce supply chain cost and ensure that the market demand of automobile parts is met in time. However, the existing automobile part demand prediction has the technical problems that the automobile part demand is influenced by the emotion of a client, but the influence factors of the emotion of the client are more, the fluctuation range is large and the prediction is difficult, so that the demand prediction precision is reduced; the method has the technical problems that the requirements of automobile parts are influenced by a plurality of factors, the feature dimension is higher, the calculation complexity is increased, the overfitting of a prediction model is easy to cause, and the capturing capacity of the prediction model on important factors is influenced; there is a technical problem that a large amount of automobile part demand data is needed for automobile part demand prediction as a basis, but the collected data is limited, so that the market demand prediction lacks sufficient relevant data support.
Disclosure of Invention
Aiming at the technical problems that the demand prediction precision is reduced because the demand of the automobile parts is influenced by the emotion of the customer, but the influence factors of the emotion of the customer are more, the fluctuation range is large and the demand is difficult to predict, an emotion mining model is adopted, the average emotion value in the observation period is calculated, the function of the sales quantity of the parts changing along with time is calculated, the effectiveness of the emotion mining model is evaluated, enterprises are helped to know the preference, satisfaction and demand of the customer for the product, and therefore the demand prediction of the automobile parts is more accurate, and the product which meets the demand of the customer is provided; aiming at the technical problems that the demand of automobile parts is influenced by a plurality of factors, the feature dimension is higher, the calculation complexity is increased, the overfitting of a prediction model is easy to cause, and the capturing capability of the prediction model on important factors is influenced, the automobile parts demand prediction model is adopted, the integrated feature selection is carried out, the comprehensive weight of the feature is calculated, an improved stacking frame is established, and the grid search cross verification is used for adjusting the parameters of a base learner; aiming at the technical problems that a large amount of automobile part demand data are needed for automobile part demand prediction, but the model is limited in collecting the automobile part demand data, so that the market demand prediction lacks sufficient relevant data support, data segmentation is adopted, a training set, a verification set and a test set of a basic learner are spliced to a meta learner, the number of samples is increased, the generalization performance of the model is improved, and a more stable prediction result is obtained.
The technical scheme adopted by the invention is as follows: the invention provides an integrated learning-based automobile part demand prediction method, which comprises the following steps:
step S1: the method comprises the steps of collecting automobile part demand data, specifically, collecting target and range of specific data, and collecting historical sales data and market trend data of parts according to the target and range, so that market demand dynamics can be known, and future demand trends can be predicted;
step S2: the data processing is specifically to carry out data cleaning, data storage and data periodic update on the automobile part demand data;
step S3: emotion data mining, namely collecting client emotion data, constructing an emotion mining model, calculating an average emotion value in an observation period, calculating a function of the sales quantity of parts changing along with time, evaluating the effectiveness of the emotion mining model, and helping enterprises to know the preference, satisfaction and demand of clients on products;
step S4: constructing an automobile part demand prediction model, specifically, obtaining potential factors influencing spare part demands through analysis of the automobile part demands, establishing an initial feature set after data cleaning and processing, acquiring a final feature set by aggregating the output of a plurality of independent feature selectors by adopting an integrated feature selection method for eliminating irrelevant and redundant features, establishing an improved stacking model, and adjusting parameters of a basic learner by using grid search cross-validation;
step S5: the method comprises the steps of optimizing a model for predicting the demands of automobile parts, particularly, carrying out data segmentation, splicing a training set, a verification set and a test set of a basic learner to a element learner, increasing the number of samples, increasing the generalization performance of the model, and obtaining a more stable prediction result.
Further, in step S1, the collecting the requirement data of the automobile parts includes the following steps:
step S11: the method comprises the steps of defining a target and a range of data collection, defining the target and the range of automobile part demand data to be collected, wherein the target and the range comprise specific part categories, time ranges and geographic ranges, and the automobile part demand data comprise part historical sales data and market trend data;
step S12: collecting historical sales data of the parts, and collecting data related to historical sales of the parts according to targets and ranges of the demand data of the automobile parts, wherein the data comprises a historical sales record, a historical sales amount and a historical sales area;
step S13: market trend data is collected, and data related to market trends is collected, including economic indicators, industry sales data, and consumer demand and preference surveys for different parts.
Further, in step S2, the data processing includes the steps of:
step S21: the data cleaning, namely cleaning and preprocessing the data required by the automobile parts, including processing the missing value, the abnormal value and the repeated value, ensuring the accuracy and the integrity of the data and improving the precision of the prediction model;
step S22: data storage, namely storing and managing the acquired automobile part demand data, and ensuring the safety and accessibility of the data;
step S23: and (3) updating data, periodically maintaining and updating the demand data of the automobile parts, ensuring timeliness and accuracy of the data, and adapting to market change and demand evolution.
Further, in step S3, the emotion data mining includes the following steps:
step S31: collecting client emotion data, and acquiring the client emotion data by crawling social media, comment platforms and questionnaire investigation modes;
step S32: carrying out emotion data processing, namely cleaning and preprocessing emotion data, and extracting emotion characteristics;
step S33: an emotion mining model is constructed, emotion analysis is carried out, and an average emotion value in an observation period is calculated, wherein the formula is as follows:
；
wherein S is t Is the average emotion value over the observation period, t is the observation period time, C is the total observation period, e is the parameter used to generate emotion analysis, C t Is customer emotion data during observation;
step S34: the function of the sales of the parts with time is calculated, and the formula is as follows:
；
wherein y is t Is a function of component sales over time t,is sales information for the first p weeks, +.>Is emotion information of the previous q weeks, i and j are indexes of p and q respectively, θ i Is the historical demand coefficient, delta j Is the emotion coefficient epsilon t Is an error term;
step S35: the effectiveness of the emotion mining model was evaluated using the following formula:
；
wherein ω is an emotion mining model validity parameter, the smaller the value of ω is, the better the model predictive ability is, N is the sample number of the emotion mining model, and tred is the estimated value of the client emotion data obtained by using the emotion mining model, that is, S t True is the actual value of the emotion data of the client and is the value obtained by data acquisition.
Further, in step S4, the building of the model for predicting the demand of the automobile parts includes the following steps:
step S41: the automobile part demand analysis, wherein the automobile part demand comes from the automobile operation and maintenance process, the part demand is related to the occurrence of automobile faults, maintenance activities, equipment age and operation conditions, and potential factors which affect the automobile part demand more comprehensively are obtained through the analysis of the three-level maintenance guarantee flow;
step S42: establishing an initial feature set, and cleaning and preprocessing data of potential factors which affect the requirements of automobile parts more comprehensively to obtain the initial feature set;
step S43: the method comprises the steps of inputting an initial feature set into an independent feature selector, aggregating the outputs of a plurality of independent feature selectors to obtain a final feature set, and selecting three basic feature selection methods to perform comprehensive weight calculation of features, wherein the comprehensive weight calculation comprises recursive feature elimination and cross verification, principal component analysis and random forest;
step S44: calculating the comprehensive weights of the features, sequencing the importance of the features into three sequences, w1, w2 and w3, defining a as a feature, obtaining the weight of the feature a through the position of the feature a in the sequence, using xi (a) to represent the position of the feature in the three methods, obtaining the comprehensive weights of the features, sequentially removing the features according to the comprehensive weights of the features, putting the features into model training and testing, determining an optimal feature subset through evaluating the performance of the model, and calculating the comprehensive weights of the features by using the following formula:
；
wherein weight (a) is the comprehensive weight of the feature, w1_ζ (a), w2_ζ (a) and w3_ζ (a) are the weights of the feature a in the three sequences, α1, α2 and α3 are the weights of the importance of the corresponding features, and α1+α2+α3=1;
step S45: establishing an improved stack model, and constructing a stack model comprising 3 basic learners and 1 element learner, wherein in the improved stack model, an optimal feature subset and an output of the basic learner are integrated in an input of the element learner;
step S46: and (3) parameter adjustment, namely adjusting parameters of the base learners by using grid search cross verification, and learning how to combine and balance the output of each base learner according to the prediction result of the base learners to obtain better performance of the automobile part demand prediction model.
Further, in step S5, the optimization of the model for predicting the demand of the automobile parts includes the following steps:
step S51: the data segmentation divides the automobile part demand data into three parts, including a training set, a verification set and a test set;
step S52: training the basic learners repeatedly, training each basic learner by using a training set, predicting on a verification set and a test set, and repeating the process for three times to obtain prediction data of the three verification sets and prediction data of the test set;
step S53: expanding a meta learner training set, vertically stacking predicted data of three verification sets, averaging the predicted data of three test sets to obtain an expanded meta learner training set, and increasing the number of samples of the meta learner training set;
step S54: expanding a meta learner verification set, horizontally splicing predicted data of the verification sets of the three basic learners to obtain an expanded meta learner verification set, so that the meta learner can perform more comprehensive verification on the verification set;
step S55: and expanding the test set of the element learner, horizontally splicing the predicted data of the test sets of the three basic learners to the test set of the element learner, reducing accidental images of the test set, and obtaining a more stable test result.
The invention provides an integrated learning-based automobile part demand prediction system, which comprises an automobile part demand data acquisition module, a data processing module, an emotion data mining module, an automobile part demand prediction model construction module and an automobile part demand prediction model optimization module, wherein the automobile part demand data acquisition module is used for acquiring the data of an automobile part;
the module for collecting the demand data of the automobile parts specifically comprises a clear data collection target and a clear data collection range, and collects historical sales data and market trend data of the parts according to the target and the range, so that the module is beneficial to knowing market demand dynamics and predicting future demand trends;
the data processing module is used for cleaning, storing and updating the data of the automobile part demand data at regular intervals;
the emotion data mining module is used for collecting emotion data of clients, constructing an emotion mining model, calculating average emotion values in an observation period, calculating functions of sales of parts changing along with time, evaluating effectiveness of the emotion mining model, and helping enterprises to know preference, satisfaction and demand of the clients on products;
the method comprises the steps of constructing an automobile part demand prediction model module, specifically, obtaining potential factors influencing spare part demands through analysis of automobile part demands, establishing an initial feature set after data cleaning and processing, acquiring a final feature set by aggregating the output of a plurality of independent feature selectors by adopting an integrated feature selection method for eliminating irrelevant and redundant features, establishing an improved stacking model, and using grid search cross-validation to adjust parameters of a base learner;
the automobile part demand prediction model optimization module is used for carrying out data segmentation, splicing a training set, a verification set and a test set of the basic learner to the element learner, increasing the number of samples, increasing the generalization performance of the model and obtaining a more stable prediction result.
The beneficial results obtained by adopting the scheme of the invention are as follows:
(1) Aiming at the technical problems that the demand of automobile parts is influenced by customer emotion, but the customer emotion influence factors are more, the fluctuation range is large and difficult to predict, and the demand prediction precision is reduced, an emotion mining model is adopted, the average emotion value in the observation period is calculated, the function of the sales quantity of the automobile parts along with the time change is calculated, the effectiveness of the emotion mining model is evaluated, enterprises are helped to know the preference, satisfaction and demand of customers on products, and therefore the demand prediction of the automobile parts is more accurate, and products which are more in line with the demands of the customers are provided;
(2) Aiming at the technical problems that the demand of automobile parts is influenced by a plurality of factors, the feature dimension is higher, the calculation complexity is increased, the overfitting of a prediction model is easy to cause, and the capturing capability of the prediction model on important factors is influenced, the automobile parts demand prediction model is adopted, the integrated feature selection is carried out, the comprehensive weight of the feature is calculated, an improved stacking frame is established, and the grid search cross verification is used for adjusting the parameters of a base learner;
(3) Aiming at the technical problems that a large amount of automobile part demand data are needed for automobile part demand prediction, but the model is limited in collecting the automobile part demand data, so that the market demand prediction lacks sufficient relevant data support, data segmentation is adopted, a training set, a verification set and a test set of a basic learner are spliced to a meta learner, the number of samples is increased, the generalization performance of the model is improved, and a more stable prediction result is obtained.
Drawings
FIG. 1 is a schematic flow chart of an integrated learning-based automobile part demand prediction method provided by the invention;
FIG. 2 is a schematic diagram of an integrated learning-based system for predicting demand for automobile parts;
FIG. 3 is a flow chart of step S3;
fig. 4 is a flow chart of step S4;
fig. 5 is a flow chart of step S5.
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention; all other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the present invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "top," "bottom," "inner," "outer," and the like indicate orientation or positional relationships based on those shown in the drawings, merely to facilitate description of the invention and simplify the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the invention.
Referring to fig. 1, the method for predicting the demand of an automobile part based on ensemble learning provided by the invention comprises the following steps:
step S1: the method comprises the steps of collecting automobile part demand data, specifically, collecting target and range of specific data, and collecting historical sales data and market trend data of parts according to the target and range, so that market demand dynamics can be known, and future demand trends can be predicted;
step S2: the data processing is specifically to carry out data cleaning, data storage and data periodic update on the automobile part demand data;
step S3: emotion data mining, namely collecting customer emotion data, constructing an emotion mining model, calculating an average emotion value in an observation period, calculating a function of the sales quantity of parts changing along with time, evaluating the effectiveness of the emotion mining model, helping enterprises to know the preference, satisfaction and demand of customers on products, so that the demand prediction of automobile parts is more accurate, and products more conforming to the demands of the customers are provided;
step S4: constructing an automobile part demand prediction model, specifically, obtaining potential factors influencing spare part demands through analysis of the automobile part demands, establishing an initial feature set after data cleaning and processing, acquiring a final feature set by aggregating the output of a plurality of independent feature selectors by adopting an integrated feature selection method for eliminating irrelevant and redundant features, establishing an improved stacking model, and adjusting parameters of a basic learner by using grid search cross-validation;
step S5: the method comprises the steps of optimizing a model for predicting the demands of automobile parts, particularly, carrying out data segmentation, splicing a training set, a verification set and a test set of a basic learner to a element learner, increasing the number of samples, increasing the generalization performance of the model, and obtaining a more stable prediction result.
In a second embodiment, referring to fig. 1, the collecting the vehicle part demand data in step S1 includes the following steps:
step S11: the method comprises the steps of defining a target and a range of data collection, defining the target and the range of automobile part demand data to be collected, wherein the target and the range comprise specific part categories, time ranges and geographic ranges, and the automobile part demand data comprise part historical sales data and market trend data;
step S12: collecting historical sales data of the parts, and collecting data related to historical sales of the parts, including historical sales records, historical sales amounts and historical sales areas, according to targets and ranges of demand data of the parts of the automobile, wherein the historical sales data of the parts are generally obtained from automobile manufacturers, parts suppliers and distributors;
step S13: market trend data is collected, and data related to market trends is collected, including economic indicators, industry sales data, and consumer demand and preference surveys for different parts.
Embodiment three, referring to fig. 1, the embodiment is based on the above embodiment, and in step S2, the data processing includes the following steps:
step S21: the data cleaning, namely cleaning and preprocessing the data required by the automobile parts, including processing the missing value, the abnormal value and the repeated value, ensuring the accuracy and the integrity of the data and improving the precision of the prediction model;
step S22: data storage, namely storing and managing the acquired automobile part demand data, and ensuring the safety and accessibility of the data;
step S23: and (3) updating data, periodically maintaining and updating the demand data of the automobile parts, ensuring timeliness and accuracy of the data, and adapting to market change and demand evolution.
In a fourth embodiment, referring to fig. 1 and 3, the emotion data mining in step S3 includes the following steps:
step S31: collecting client emotion data, and acquiring the client emotion data by crawling social media, comment platforms and questionnaire investigation modes;
step S32: carrying out emotion data processing, namely cleaning and preprocessing emotion data, and extracting emotion characteristics;
step S33: an emotion mining model is constructed, emotion analysis is carried out, and an average emotion value in an observation period is calculated, wherein the formula is as follows:
；
wherein y is t Is a function of the variation of the sales of the component with time t, p and q are specific numbers of cycles, i and j are indices of p and q, respectively, θ i Is the historical demand coefficient, delta j Is the emotion coefficient epsilon t Is an error term;
step S34: the function of the sales of the parts with time is calculated, and the formula is as follows:
；
wherein y is t Is a function of component sales over time t,is sales information for the first p weeks, +.>Is emotion information of the previous q weeks, i and j are indexes of p and q respectively, θ i Is the historical demand coefficient, delta j Is the emotion coefficient epsilon t Is an error term;
step S35: the effectiveness of the emotion mining model was evaluated using the following formula:
；
wherein ω is an emotion mining model validity parameter, the smaller the value of ω, the better the model predictive ability, N is the sample number of the emotion mining model, tred is the emotion mining modelThe obtained estimated value of the emotion data of the client is S t True is the actual value of the emotion data of the client and is the value obtained by data acquisition.
Through executing the operation, the emotion mining model is adopted, the average emotion value in the observation period is calculated, the function of the change of the sales quantity of the parts along with time is calculated, the effectiveness of the emotion mining model is evaluated, enterprises are helped to know the preference, satisfaction and demand of customers on products, accordingly, the demand prediction of the automobile parts is more accurate, products which are more in line with the demands of the customers are provided, the technical problems that the demands of the automobile parts are influenced by the emotion of the customers, but the influence factors of the emotion of the customers are more, the fluctuation range is large and difficult to predict, and the demand prediction precision is reduced are solved.
Fifth embodiment, referring to fig. 1 and 4, the method for constructing a model for predicting the demand of an automobile part in step S4 includes the following steps:
step S41: the automobile part demand analysis, wherein the automobile part demand comes from the automobile operation and maintenance process, the part demand is related to the occurrence of automobile faults, maintenance activities, equipment age and operation conditions, and potential factors which affect the automobile part demand more comprehensively are obtained through the analysis of the three-level maintenance guarantee flow;
step S42: establishing an initial feature set, and cleaning and preprocessing data of potential factors which affect the requirements of automobile parts more comprehensively to obtain the initial feature set;
step S43: the method comprises the steps of inputting an initial feature set into an independent feature selector, aggregating the outputs of a plurality of independent feature selectors to obtain a final feature set, and selecting three basic feature selection methods to perform comprehensive weight calculation of features, wherein the comprehensive weight calculation comprises recursive feature elimination and cross verification, principal component analysis and random forest;
step S44: calculating the comprehensive weights of the features, sequencing the importance of the features into three sequences, w1, w2 and w3, defining a as a feature, obtaining the weight of the feature a through the position of the feature a in the sequence, using xi (a) to represent the position of the feature in the three methods, obtaining the comprehensive weights of the features, sequentially removing the features according to the comprehensive weights of the features, putting the features into model training and testing, determining an optimal feature subset through evaluating the performance of the model, and calculating the comprehensive weights of the features by using the following formula:
；
wherein weight (a) is the comprehensive weight of the feature, w1_ζ (a), w2_ζ (a) and w3_ζ (a) are the weights of the feature a in the three sequences, α1, α2 and α3 are the weights of the importance of the corresponding features, and α1+α2+α3=1;
step S45: establishing an improved stack model, and constructing a stack model comprising 3 basic learners and 1 element learner, wherein in the improved stack model, an optimal feature subset and an output of the basic learner are integrated in an input of the element learner;
step S46: and (3) parameter adjustment, namely adjusting parameters of the base learners by using grid search cross verification, and learning how to combine and balance the output of each base learner according to the prediction result of the base learners to obtain better performance of the automobile part demand prediction model.
Through executing the operation, the automobile part demand prediction model is adopted to perform integrated feature selection, comprehensive weights of features are calculated, an improved stacking frame is established, grid search cross verification is used to adjust parameters of a base learner, and the technical problems that automobile part demands are affected by a plurality of factors, feature dimensions are high, calculation complexity is increased, the prediction model is easy to fit excessively, and capturing capacity of the prediction model on important factors is affected are solved.
Embodiment six, referring to fig. 1 and 5, based on the above embodiment, in step S5, the optimization of the model for predicting the demand of automobile parts includes the following steps:
step S51: the data segmentation divides the automobile part demand data into three parts, including a training set, a verification set and a test set;
step S52: training the basic learners repeatedly, training each basic learner by using a training set, predicting on a verification set and a test set, and repeating the process for three times to obtain prediction data of the three verification sets and prediction data of the test set;
step S53: expanding a meta learner training set, vertically stacking predicted data of three verification sets, averaging the predicted data of three test sets to obtain an expanded meta learner training set, and increasing the number of samples of the meta learner training set;
step S54: expanding a meta learner verification set, horizontally splicing predicted data of the verification sets of the three basic learners to obtain an expanded meta learner verification set, so that the meta learner can perform more comprehensive verification on the verification set;
step S55: and expanding the test set of the element learner, horizontally splicing the predicted data of the test sets of the three basic learners to the test set of the element learner, reducing accidental images of the test set, and obtaining a more stable test result.
By executing the operation, the data segmentation is adopted, the training set, the verification set and the test set of the basic learner are spliced to the element learner, the number of samples is increased, the generalization performance of the model is increased, a more stable prediction result is obtained, and the technical problem that a large amount of automobile part demand data is needed for automobile part demand prediction as a basis, but the model is limited in collecting the automobile part demand data, so that the market demand prediction lacks sufficient relevant data support is solved.
An embodiment seven, referring to fig. 2, based on the embodiment, the invention provides an integrated learning-based automobile part demand prediction system, which comprises an automobile part demand data acquisition module, a data processing module, an emotion data mining module, an automobile part demand prediction model building module and an automobile part demand prediction model optimizing module;
the module for collecting the demand data of the automobile parts specifically comprises a clear data collection target and a clear data collection range, and collects historical sales data and market trend data of the parts according to the target and the range, so that the module is beneficial to knowing market demand dynamics and predicting future demand trends;
the data processing module is used for cleaning, storing and updating the data of the automobile part demand data at regular intervals;
the emotion data mining module is used for collecting emotion data of a client, constructing an emotion mining model, calculating an average emotion value in an observation period, calculating a function of the sales quantity of parts changing along with time, evaluating the effectiveness of the emotion mining model, helping enterprises to know the preference, satisfaction and demand of the client for products, and further enabling the demand prediction of automobile parts to be more accurate and providing products more conforming to the demands of the client;
the method comprises the steps of constructing an automobile part demand prediction model module, specifically, obtaining potential factors influencing spare part demands through analysis of automobile part demands, establishing an initial feature set after data cleaning and processing, acquiring a final feature set by aggregating the output of a plurality of independent feature selectors by adopting an integrated feature selection method for eliminating irrelevant and redundant features, establishing an improved stacking model, and using grid search cross-validation to adjust parameters of a base learner;
the automobile part demand prediction model optimization module is used for carrying out data segmentation, splicing a training set, a verification set and a test set of the basic learner to the element learner, increasing the number of samples, increasing the generalization performance of the model and obtaining a more stable prediction result.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
The invention and its embodiments have been described above with no limitation, and the actual construction is not limited to the embodiments of the invention as shown in the drawings. In summary, if one of ordinary skill in the art is informed by this disclosure, a structural manner and an embodiment similar to the technical solution should not be creatively devised without departing from the gist of the present invention.
Claims (7)
1. An automobile part demand prediction method based on ensemble learning is characterized by comprising the following steps: the method comprises the following steps:
step S1: collecting automobile part demand data;
step S2: the data processing is specifically to clean, store and update the automobile part demand data;
step S3: the emotion data mining is specifically implemented by constructing an emotion mining model, calculating functions of average emotion values and part sales in an observation period along with time change, and evaluating the effectiveness of the emotion mining model;
step S4: constructing an automobile part demand prediction model, specifically adopting an integrated feature selection method to calculate comprehensive weights of features, constructing an improved stacking model, and using grid search cross validation to adjust parameters of a base learner;
step S5: optimizing a vehicle part demand prediction model, namely performing data segmentation, and splicing a training set, a verification set and a test set of a basic learner to the basic learner;
in step S3, the emotion data mining includes the following steps:
step S31: collecting client emotion data, and acquiring the client emotion data by crawling social media, comment platforms and questionnaire investigation modes;
step S32: carrying out emotion data processing, namely cleaning and preprocessing emotion data, and extracting emotion characteristics;
step S33: an emotion mining model is constructed, emotion analysis is carried out, and an average emotion value in an observation period is calculated, wherein the formula is as follows:
；
wherein S is t Is the average emotion value over the observation period, t is the observation period time, C is the total observation period, e is the parameter used to generate emotion analysis, C t Is customer emotion data during observation;
step S34: the function of the sales of the parts with time is calculated, and the formula is as follows:
；
wherein y is t Is a function of component sales over time t,is sales information for the first p weeks, +.>Is emotion information of the previous q weeks, i and j are indexes of p and q respectively, θ i Is the historical demand coefficient, delta j Is the emotion coefficient epsilon t Is an error term;
step S35: the effectiveness of the emotion mining model was evaluated using the following formula:
；
wherein ω is an emotion mining model validity parameter, the smaller the value of ω is, the better the model predictive ability is, N is the sample number of the emotion mining model, and tred is the estimated value of the client emotion data obtained by using the emotion mining model, that is, S t True is customer emotion dataIs the actual value obtained by data acquisition.
2. The method for predicting the demand of automobile parts based on ensemble learning according to claim 1, wherein the method comprises the steps of: in step S4, the building of the automobile part demand prediction model includes the following steps:
step S41: the automobile part demand analysis is carried out, the part demand is related to fault occurrence, maintenance activities, equipment age and operation conditions, and potential factors which influence the automobile part demand more comprehensively are obtained through the analysis of three-level maintenance and guarantee flows;
step S42: establishing an initial feature set, and cleaning and preprocessing data of potential factors which affect the requirements of automobile parts more comprehensively to obtain the initial feature set;
step S43: the method comprises the steps of inputting an initial feature set into an independent feature selector, aggregating the outputs of a plurality of independent feature selectors to obtain a final feature set, and selecting three basic feature selection methods to perform comprehensive weight calculation of features, wherein the comprehensive weight calculation comprises recursive feature elimination and cross verification, principal component analysis and random forest;
step S44: calculating the comprehensive weights of the features, sequencing the importance of the features into three sequences, w1, w2 and w3, defining a as a feature, obtaining the weight of the feature a through the position of the feature a in the sequence, using xi (a) to represent the position of the feature in the three methods, obtaining the comprehensive weights of the features, sequentially removing the features according to the comprehensive weights of the features, putting the features into model training and testing, determining an optimal feature subset through evaluating the performance of the model, and calculating the comprehensive weights of the features by using the following formula:
；
wherein weight (a) is the comprehensive weight of the feature, w1_ζ (a), w2_ζ (a) and w3_ζ (a) are the weights of the feature a in the three sequences, α1, α2 and α3 are the weights of the importance of the corresponding features, and α1+α2+α3=1;
step S45: establishing an improved stack model, and constructing a stack model comprising 3 basic learners and 1 element learner, wherein in the improved stack model, an optimal feature subset and an output of the basic learner are integrated in an input of the element learner;
step S46: and (3) parameter adjustment, namely adjusting parameters of the base learners by using grid search cross verification, and learning how to combine and balance the output of each base learner according to the prediction result of the base learners to obtain better performance of the automobile part demand prediction model.
3. The method for predicting the demand of automobile parts based on ensemble learning according to claim 1, wherein the method comprises the steps of: in step S5, the performance optimization of the automobile part demand prediction model includes the following steps:
step S51: the data segmentation divides the automobile part demand data into three parts, including a training set, a verification set and a test set;
step S52: training the basic learners repeatedly, training each basic learner by using a training set, predicting on a verification set and a test set, and repeating the process for three times to obtain prediction data of the three verification sets and prediction data of the test set;
step S53: expanding a meta learner training set, vertically stacking the predicted data of the three verification sets, and averaging the predicted data of the three test sets to obtain an expanded meta learner training set;
step S54: expanding a meta learner verification set, and horizontally splicing predicted data of the verification sets of the three base learners to obtain an expanded meta learner verification set;
step S55: the test set of the element learner is expanded, and the predicted data levels of the test sets of the three base learners are spliced to the test set of the element learner.
4. The method for predicting the demand of automobile parts based on ensemble learning according to claim 1, wherein the method comprises the steps of: in step S1, the collecting the demand data of the automobile parts includes the following steps:
step S11: defining a target and a range of data collection, and defining the target and the range of the requirement data of the automobile parts to be collected;
step S12: collecting historical sales data of the parts, and collecting data related to historical sales of the parts according to the target and the range of the demand data of the automobile parts;
step S13: market trend data is collected, and data related to market trends is collected, including economic indicators, industry sales data, and consumer demand and preference surveys for different parts.
5. The method for predicting the demand of automobile parts based on ensemble learning according to claim 1, wherein the method comprises the steps of: in step S2, the data processing includes the steps of:
step S21: data cleaning, namely cleaning and preprocessing the data required by the automobile parts;
step S22: storing and managing the acquired automobile part demand data;
step S23: and (5) updating data, and periodically maintaining and updating the demand data of the automobile parts.
6. An integrated learning-based automobile part demand prediction system for implementing the integrated learning-based automobile part demand prediction method as set forth in any one of claims 1 to 5, wherein: the system comprises a module for collecting automobile part demand data, a data processing module, an emotion data mining module, a module for constructing an automobile part demand prediction model and an optimization module for constructing an automobile part demand prediction model.
7. The vehicle component demand prediction system based on ensemble learning of claim 6, wherein: the module for collecting the demand data of the automobile parts specifically comprises a specific data collection target and range, and historical sales data and market trend data of the parts are collected according to the target and range;
the data processing module is used for cleaning, storing and updating the data of the automobile part demand data at regular intervals;
the emotion data mining module is used for collecting client emotion data, constructing an emotion mining model, calculating an average emotion value in an observation period, calculating a function of the sales quantity of parts changing along with time, and evaluating the effectiveness of the emotion mining model;
the method comprises the steps of constructing an automobile part demand prediction model module, specifically, obtaining potential factors influencing spare part demands through analysis of automobile part demands, establishing an initial feature set after data cleaning and processing, acquiring a final feature set by aggregating the output of a plurality of independent feature selectors by adopting an integrated feature selection method, establishing an improved stacking model, and adjusting parameters of a basic learner by using grid search cross verification;
the automobile part demand prediction model optimization module is used for carrying out data segmentation, and splicing a training set, a verification set and a test set of the basic learner to the element learner to obtain a more stable prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311502410.1A CN117236527B (en) | 2023-11-13 | 2023-11-13 | Automobile part demand prediction method and system based on ensemble learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311502410.1A CN117236527B (en) | 2023-11-13 | 2023-11-13 | Automobile part demand prediction method and system based on ensemble learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117236527A true CN117236527A (en) | 2023-12-15 |
| CN117236527B CN117236527B (en) | 2024-02-06 |
Family
ID=89082963
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311502410.1A Active CN117236527B (en) | 2023-11-13 | 2023-11-13 | Automobile part demand prediction method and system based on ensemble learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117236527B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109903061A (en) * | 2017-12-07 | 2019-06-18 | 厦门雅迅网络股份有限公司 | A kind of automobile parts needing forecasting method, terminal device and storage medium |
| CN112258251A (en) * | 2020-11-18 | 2021-01-22 | 北京理工大学 | Grey correlation-based integrated learning prediction method and system for electric vehicle battery replacement demand |
| CN112949948A (en) * | 2021-04-28 | 2021-06-11 | 北京理工大学 | Integrated learning method and system for electric vehicle power conversion demand interval prediction in time-sharing mode |
| KR20220008492A (en) * | 2020-07-14 | 2022-01-21 | 주식회사 엠제이비전테크 | Method for providing deep learning model based vehicle part testing service |
| CN114782110A (en) * | 2022-05-10 | 2022-07-22 | 中国银行股份有限公司 | Demand mining method and system based on logistic regression two-classification and JMTS |
| CN115438849A (en) * | 2022-08-29 | 2022-12-06 | 北京航空航天大学 | Demand prediction method for subsequent spare parts of equipment based on ensemble learning |
| CN116307652A (en) * | 2023-05-25 | 2023-06-23 | 华北电力大学 | Artificial intelligent resource allocation method for intelligent power grid |
| CN116843156A (en) * | 2023-07-28 | 2023-10-03 | 西华大学 | Supply chain management and control method for assembled building components |
-
2023
- 2023-11-13 CN CN202311502410.1A patent/CN117236527B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109903061A (en) * | 2017-12-07 | 2019-06-18 | 厦门雅迅网络股份有限公司 | A kind of automobile parts needing forecasting method, terminal device and storage medium |
| KR20220008492A (en) * | 2020-07-14 | 2022-01-21 | 주식회사 엠제이비전테크 | Method for providing deep learning model based vehicle part testing service |
| CN112258251A (en) * | 2020-11-18 | 2021-01-22 | 北京理工大学 | Grey correlation-based integrated learning prediction method and system for electric vehicle battery replacement demand |
| CN112949948A (en) * | 2021-04-28 | 2021-06-11 | 北京理工大学 | Integrated learning method and system for electric vehicle power conversion demand interval prediction in time-sharing mode |
| CN114782110A (en) * | 2022-05-10 | 2022-07-22 | 中国银行股份有限公司 | Demand mining method and system based on logistic regression two-classification and JMTS |
| CN115438849A (en) * | 2022-08-29 | 2022-12-06 | 北京航空航天大学 | Demand prediction method for subsequent spare parts of equipment based on ensemble learning |
| CN116307652A (en) * | 2023-05-25 | 2023-06-23 | 华北电力大学 | Artificial intelligent resource allocation method for intelligent power grid |
| CN116843156A (en) * | 2023-07-28 | 2023-10-03 | 西华大学 | Supply chain management and control method for assembled building components |
Non-Patent Citations (1)
| Title |
|---|
| 吴春华: "基于需求预测的S公司汽车零部件采购计划优化研究", pages 1 - 71 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117236527B (en) | 2024-02-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109784806B (en) | Supply chain control method, system and storage medium | |
| Van Horenbeek et al. | Quantifying the added value of an imperfectly performing condition monitoring system—Application to a wind turbine gearbox | |
| JP2020500420A (en) | Yield prediction system and method for machine learning based semiconductor manufacturing | |
| WO2010138452A1 (en) | Apparatus and method for model quality estimation and model adaptation in multivariable process control | |
| CN108460521A (en) | The recommendation method and system of the audit target | |
| Ungermann et al. | Data analytics for manufacturing systems–a data-driven approach for process optimization | |
| CN116739669B (en) | System and method for monitoring ocpx advertisements in real time | |
| JP7156975B2 (en) | Management evaluation device, management evaluation method, and program | |
| CN117236527B (en) | Automobile part demand prediction method and system based on ensemble learning | |
| CN114529204A (en) | Modularized production process control and management method and system | |
| CN116703455B (en) | Medicine data sales prediction method and system based on time series hybrid model | |
| US7359842B1 (en) | Soft sensor device and device for evaluating the same | |
| CN111950775A (en) | Economic operation monitoring method based on big data | |
| CN107480703A (en) | Transaction fault detection method and device | |
| CN115688016A (en) | Integrated method for intelligent operation and maintenance of large database | |
| Babakmehr et al. | Data‐driven framework for warranty claims forecasting with an application for automotive components | |
| CN116149895A (en) | Big data cluster performance prediction method and device and computer equipment | |
| CN114252794B (en) | Method and device for predicting residual life of disassembled intelligent ammeter | |
| CN115759263A (en) | Strategy effect evaluation method and device based on cause and effect inference | |
| CN114048592A (en) | Finish rolling whole-flow distributed operation performance evaluation and non-optimal reason tracing method | |
| CN111798069A (en) | Machine learning-based complex equipment spare part management method and system | |
| CN112801388A (en) | Power load prediction method and system based on nonlinear time series algorithm | |
| Wang | Modelling condition monitoring inspection using the delay-time concept | |
| CN112070415B (en) | Scenarioized baseline and model implementation method established based on software quantitative management process | |
| JP2021082207A (en) | Waste carry-in amount prediction device, waste carry-in amount prediction method, and waste carry-in amount prediction program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |