CN110097287B - Group portrait method for logistic drivers - Google Patents

Group portrait method for logistic drivers Download PDF

Info

Publication number
CN110097287B
CN110097287B CN201910376458.XA CN201910376458A CN110097287B CN 110097287 B CN110097287 B CN 110097287B CN 201910376458 A CN201910376458 A CN 201910376458A CN 110097287 B CN110097287 B CN 110097287B
Authority
CN
China
Prior art keywords
driver
label
similarity
order
quantitative
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.)
Active
Application number
CN201910376458.XA
Other languages
Chinese (zh)
Other versions
CN110097287A (en
Inventor
詹会兰
庭治宏
施甘图
李贞昊
赵亮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Lahuobao Network Technology Co ltd
Hongtu Logistics Co ltd
Original Assignee
Lahuobao Network Technology Co ltd
Hongtu Logistics Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Lahuobao Network Technology Co ltd, Hongtu Logistics Co ltd filed Critical Lahuobao Network Technology Co ltd
Priority to CN201910376458.XA priority Critical patent/CN110097287B/en
Publication of CN110097287A publication Critical patent/CN110097287A/en
Application granted granted Critical
Publication of CN110097287B publication Critical patent/CN110097287B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2465—Query processing support for facilitating data mining operations in structured databases
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06311—Scheduling, planning or task assignment for a person or group
    • G06Q10/063114—Status monitoring or status determination for a person or group
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06315—Needs-based resource requirements planning or analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • G06Q10/083—Shipping
    • G06Q10/0835—Relationships between shipper or supplier and carriers

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Theoretical Computer Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Tourism & Hospitality (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Databases & Information Systems (AREA)
  • Fuzzy Systems (AREA)
  • Mathematical Physics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Software Systems (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a group portrait method of a logistics driver, which is applied to the field of logistics dispatching and aims at solving the problems that the current driver portrait lacks dynamic information and is not in line with the actual logistics situation; according to the method, the multi-dimensional information of drivers of logistics vehicles is integrated, the similarity of any two drivers on each label is calculated, and then the similarities on all the labels are weighted and summed to obtain the similarity between the two drivers; then clustering the drivers according to the similarity, thereby generating a group driver portrait; when the similarity of any two drivers on the labels is calculated, particularly the calculation of the similarity of the time-route labels is considered, and the service dynamic information is fully utilized, so that the finally obtained driver group portrait is dynamic and more accords with the actual situation.

Description

Group portrait method for logistic drivers
Technical Field
The invention belongs to the field of logistics scheduling, and particularly relates to a logistics driver image technology.
Background
With the explosive development of the logistics industry, highly information-based third-party logistics platforms are in dispute, and the logistics service providers have certain manpower and vehicle resources and are popularized and served to a large number of medium-sized and small logistics enterprises and other large, medium-sized and small enterprises. The service patterns provided by these platforms are: and the customer with the logistics distribution service requirement goes to a third-party logistics platform to submit a logistics vehicle scheduling request, and then the platform returns the recommended vehicle set.
In the third-party logistics platform recommendation applications, in order to enable recommendation results to meet actual requirements of customers, portrait modeling needs to be performed on drivers, so that important information such as behavior habits and order taking preferences of the drivers is analyzed, and a large amount of data bases are provided. The driver image modeling is labeling of driver information so as to provide more accurate information for a later recommendation algorithm.
The driver portrait can realize refined summarization of the driver by labeling the driver, thereby facilitating the quick understanding of people and being well processed by a computer. In practice, there is a correlation between different drivers, so that not only a single driver representation should be made for analyzing the target driver, but also a correlation analysis between drivers, i.e. a group driver representation analysis, should be included.
The group driver portrait is a sketch of the real characteristics of the target driver group, is a comprehensive prototype of the group driver and represents a group of real drivers. Analyzing the group driver portrait, classifying drivers with the same characteristics into the same group by a clustering mode according to different evaluation dimensions. A plurality of drivers on the platform are divided into a plurality of driver groups to represent all drivers owned by the platform, so that the drivers can visually feel, and the online drivers can conveniently communicate and manage.
For example, patent application publication No. CN108985819A provides a group portrait method, but it does not reflect dynamic driver portrait in logistics scheduling, and is not suitable for actual logistics situation.
Disclosure of Invention
In order to solve the technical problem, the invention provides a group portrait method of logistics drivers, which makes full use of business dynamic information to make the group portrait dynamic and more suitable for practical situations.
The technical scheme adopted by the invention is as follows: a group portrait method for logistic drivers comprises the following steps:
s1, qualitative driver image including the granularity and qualitative label;
s2, driver quantitative portrait including quantitative labels for determining driver dimension, vehicle dimension and customer dimension; the driver dimension quantitative label comprises at least a driver order taking time-route label;
s3, converting the qualitative labels into quantitative labels, and calculating the similarity of two drivers on each quantitative label to obtain the driver portrait similarity;
s4, clustering the driver portrait according to the similarity of the driver portrait;
and S5, generating and updating the group driver portrait according to the clustering result of the step S4.
Further, step S3 specifically includes the following steps:
s31, converting the qualitative labels into quantitative labels, normalizing the quantitative labels of continuous value types, and performing binary variation quantization on the quantitative labels of discrete category values;
s32, calculating the similarity of the two driver images processed in the step S31 on a certain quantitative label;
and S33, summing the similarity of all quantitative labels of the two drivers to obtain the similarity between the two drivers.
Further, step S32 includes: quantitative label similarity calculation for some binary variable:
Figure GDA0003052365050000021
wherein, sim (U)i,Uj) Driver representation UiAnd driver portrait UjSimilarity on this binary variable quantitative Label, label (U)i) Driver representation UiThe value on the quantitative Label, label (U)j) Driver representation UjA value on the quantitative label;
quantitative label similarity calculation for some non-binary variable:
Figure GDA0003052365050000022
wherein, sim (U)i,Uj) Driver representation UiAnd driver portrait UjIn the second placeSimilarity on Meta-variant quantitative labels, labelk(Ui) Driver representation UiValue on the kth quantitative Label, labelk(Uj) Driver representation UjThe value on the kth quantitative label.
Further, step S32 includes: the similarity calculation process of the order taking time-route of the driver is as follows:
a1, the driver order-receiving time-route type matrix is:
Figure GDA0003052365050000023
namely: TDL ═ xij}m×n，(1≤i≤m,1≤j≤n)
Where m represents the number of drivers, n route categories, tiDenotes a time period, xijIs shown over a time period tiThe order route of the inner driver i for taking orders is j times;
a2, obtaining the similarity according to the driver order-receiving time-route type matrix:
Figure GDA0003052365050000031
wherein L is a route type, L is a set of route types,
Figure GDA0003052365050000032
represents tiThe number of times that driver d takes an order of type l in the time slot is equivalent to x in the TDL matrixij；
A3, calculating time period tiAnd time period tjThe interval therebetween:
Figure GDA0003052365050000033
a4, calculating the initial time similarity:
Figure GDA0003052365050000034
a5, smoothing the time-route matrix of driver order receiving if driver diAt tkOrders of type l for the route have not been accepted during the time period, i.e.
Figure GDA0003052365050000035
If 0, then use driver diThe number of times of order with type of single route as l in other time periods and tkSimilarity pair of
Figure GDA0003052365050000036
Carrying out prediction; the calculation is as follows:
Figure GDA0003052365050000037
a6, calculating the similarity of the driver order taking time-route according to the data after the smoothing treatment as follows:
Figure GDA0003052365050000038
further, the driver dimension label of step S2 includes: cleaning basic attributes of a driver in a logistics driver database owned by a logistics service provider, wherein the basic attributes comprise name, age, gender and contact information;
the driver behavior information including order receiving times, freight times and freight mileage is excavated by jointly checking the driver data sheet and the order data sheet;
and (4) jointly checking a driver data table, an order table, a client table and a cargo table, and mining the interest preference of the driver, including the order receiving time preference, the cargo route preference, the transaction client preference and the order receiving price preference.
Further, the vehicle dimension label of step S2 includes: obtaining customer characteristics including names, industries and contact information in a historical transaction database of a driver of the logistics vehicle;
checking historical transaction data and an order data table of a driver to obtain customer behaviors including order times and cash settlement times;
the driver data table, the goods data table, and the order data table are co-reviewed to obtain customer interest preferences including order time preferences, goods characteristics preferences, driver trading preferences, order price preferences, and order route preferences.
Further, the customer dimension label of step S2 includes: obtaining customer characteristics including names, industries and contact information in a historical transaction database of a driver of the logistics vehicle;
checking historical transaction data and an order data table of a driver to obtain customer behaviors including order times and cash settlement times;
the driver data table, the goods data table, and the order data table are co-reviewed to obtain customer interest preferences including order time preferences, goods characteristics preferences, driver trading preferences, order price preferences, and order route preferences.
The invention has the beneficial effects that: according to the method, a single logistics driver image is obtained through big data processing, on the basis, the similarity of each quantitative label in the logistics driver image is calculated, and then the group image of the logistics driver is obtained through a clustering method. For a logistics service provider with certain manpower and vehicle resources, the logistics drivers are classified into a plurality of logistics driver image classes with small class difference and large class-to-class difference through the method of the invention, and the structural condition of the manpower transport capacity service capability of the logistics service provider can be reflected from multiple directions; the logistics driver images are clustered in multiple layers, so that the logistics drivers are depicted more comprehensively, the success rate of subsequent accurate recommendation is improved, and the service quality and the competitiveness of the platform are improved; in addition, the group images of the logistics drivers enable the logistics drivers to be clustered according to the images, and offline communication between the drivers of the same group images is facilitated; in addition, the invention considers the calculation of time-route similarity, and incorporates the calculation into the labels required by clustering, and fully utilizes the service dynamic information, so that the group portrait is dynamic and more accords with the actual situation.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a schematic view of a driver image structure.
Detailed Description
In order to facilitate understanding of the technical contents of the present invention, the following technical terms are defined:
1. qualitative sketch
The extent of the range is determined for the image.
2. Qualitative label
Tags obtained by means of questionnaires.
3. Quantitative portrait
Quantitative labels are determined for attributes, interests, behaviors, etc. of the representation.
4. Quantitative label
Directly computable labels taken from the driver dimension, vehicle dimension, and customer dimension.
In order to facilitate the understanding of the technical contents of the present invention by those skilled in the art, the present invention will be further explained with reference to the accompanying drawings.
Fig. 1 shows a flowchart of an embodiment of the present invention, and the method for group portrayal of logistic drivers of the present invention includes:
s1, qualitative driver image including the granularity and qualitative label;
s2, driver quantitative portrait including quantitative labels for determining driver dimension, vehicle dimension and customer dimension; the driver dimension quantitative label comprises at least a driver order taking time-route label;
s3, converting the qualitative labels into quantitative labels, and calculating the similarity of two drivers on each quantitative label to obtain the driver portrait similarity;
s4, clustering the driver portrait according to the similarity of the driver portrait;
and S5, generating and updating the group driver portrait according to the clustering result of the step S4.
The driver portrait mainly comprises a quantitative label and a qualitative label, the specific driver portrait structure is shown in figure 2, and the specific label is already in the step
Step S1 specifically includes: in the modeling process of qualitative driver portraits, the granularity of the driver portraits, namely the degree to which the driver portraits should be refined, is mainly considered. The method adopts a questionnaire survey mode to know the transaction scene, the order content and the like of a driver and determine the label and the image granularity of the driver image.
The three dimensions of step S2 are specifically:
1) the driver dimension is that basic attributes of the driver, including name, age, gender and contact information, are cleaned in a logistics driver database owned by a logistics service provider; the driver behavior information including order receiving times, freight times and freight mileage is excavated by jointly checking the driver data sheet and the order data sheet; and (4) jointly checking a driver data table, an order table, a client table and a cargo table, and mining the interest preference of the driver, including the order receiving time preference, the cargo route preference, the transaction client preference and the order receiving price preference.
2) The vehicle dimension is that vehicle attributes including length, width, height and load limit are cleaned in a logistics vehicle database owned by a logistics service provider; and acquiring the real-time position of the vehicle through a third-party real-time positioning service module interface.
3) The method comprises the steps that customer dimensions are obtained in a historical transaction database of a driver of the logistics vehicle, and customer characteristics including names, industries and contact ways are obtained; checking historical transaction data and an order data table of a driver to obtain customer behaviors including order times and cash settlement times; the driver data table, the goods data table, and the order data table are co-reviewed to obtain customer interest preferences including order time preferences, goods characteristics preferences, driver trading preferences, order price preferences, and order route preferences.
Step S3 specifically includes:
and preprocessing the obtained value of each label of the driver. The quantitative label value of the continuous value type is normalized, and the discrete type value is subjected to binary variation quantization processing. Then, the qualitative labels are converted into quantitative labels, the similarity of each quantitative label of the driver portrait is calculated, finally, the similarity of each quantitative label is weighted and summed to obtain the similarity between the driver portraits, the driver portraits with high similarity are classified into one class, and a precondition is provided for developing driver portrait clustering.
The calculation process of the similarity between the driver quantitative images comprises the following steps:
s31, normalization of the data of the quantitative label:
Figure GDA0003052365050000061
wherein X represents a value before tag conversion, Y represents a value after tag conversion, Xmax、XminRespectively representing a maximum value and a minimum value identifying the value to be converted.
S32, calculating the similarity of quantitative labels of certain binary variables:
Figure GDA0003052365050000062
wherein, sim (U)i,Uj) Driver representation UiAnd driver portrait UjSimilarity on this binary variable quantitative Label, label (U)i) Driver representation UiThe value on the quantitative Label, label (U)j) Driver representation UjThe value on the quantitative label. In this case, the quantitative label is a binary variable, i.e., takes on values of only 0 and 1.
S33, calculating the similarity of quantitative labels of certain non-binary variables:
Figure GDA0003052365050000063
wherein, sim (U)i,Uj) Driver representation UiAnd driver portrait UjSimilarity on the non-binary quantitative Label, labelk(Ui) Driver representation UiValue on the kth quantitative Label, labelk(Uj) Driver representation UjAt the k quantitative determinationThe value on the tag.
And (3) calculating cosine similarity by considering the statistical characteristics of the label values of the driver portrait:
Figure GDA0003052365050000064
wherein, sim (U)i,Uj) Driver representation UiAnd driver portrait UjSimilarity on the non-binary quantitative Label, labelk(Ui) Driver representation UiValue on the kth quantitative Label, labelk(Uj) Driver representation UjThe value on the k-th quantitative label,
Figure GDA0003052365050000071
driver representation UiThe average value of the image of the belonging population on the kth quantitative label,
Figure GDA0003052365050000072
driver representation UjMean of the images of the population at the kth quantitative label.
S34, calculation of similarity of a set (k) of quantitative labels (i.e. all quantitative labels):
Figure GDA0003052365050000073
wherein, wkWeight, w, representing the kth quantitative labelkObtained by sequential binary comparative quantization based on human empirical knowledge, sim (label)k(ci),labelk(cj) A picture c) representing the driveriAnd a driver figure cjSimilarity on kth quantitative label.
The invention especially considers the similarity of driver order-receiving time-route, and the calculation process is as follows:
suppose there are m drivers, there are n routes, at tiIn the time period, a time driver-route can be formedThe type matrix is:
Figure GDA0003052365050000074
namely: TDL ═ xij}m×n，(1≤i≤m,1≤j≤n)
Wherein x isijIs shown over a time period tiAnd the order route of the driver i for taking the order is j times.
And obtaining similarity according to the driver order receiving time-route type matrix:
Figure GDA0003052365050000075
wherein L is a route type, L is a set of route types,
Figure GDA0003052365050000076
represents tiThe number of times that driver d takes an order of type l in the time slot is equivalent to x in the TDL matrixij。
Calculating the time period tiAnd time period tjThe interval therebetween:
Figure GDA0003052365050000077
calculating the initial time similarity:
Figure GDA0003052365050000081
smoothing the time-route matrix of driver order if driver diAt tkOrders of type l for the route have not been accepted during the time period, i.e.
Figure GDA0003052365050000082
If 0, then use driver diNumber of times of order of type l of single route in other time periods andother time periods and tkSimilarity pair of
Figure GDA0003052365050000083
And (6) performing prediction. The calculation is as follows:
Figure GDA0003052365050000084
according to the data after the smoothing processing, calculating the similarity of the driver order taking time-route:
Figure GDA0003052365050000085
step S4 specifically includes: calculating mutual entropy values among the driver images and a total entropy value of a single driver image based on the similarity, and finally clustering the driver images by adopting a k-means clustering algorithm based on k driver images with the largest entropy values as a clustering center to obtain a group image of the driver images. And clustering by adopting a K-means algorithm, setting K as t, taking the data of the central node clustered in the last iteration as a cluster driver portrait of the cluster, and dividing the driver portrait into t classes. The specific clustering algorithm is the prior art, and the invention is not elaborated herein.
Step S5 specifically includes: the historical data of the driver representation is calculated and intermediate values of different granularities are stored in a database. The updating of the driver portrait is realized by adopting a sliding window filtering algorithm. When the driver picture database is updated, the time window is moved, the old candidate set is deleted, and the newly added candidate item set is added. And then calculating to obtain a new driver portrait based on the intermediate values and the newly added data stored in the database, and storing the updated intermediate values with different granularities into the database. It will be understood by those skilled in the art that the intermediate values of different granularities mentioned herein include data generated throughout the rendering process, such as: the values of the labels, the number of clusters generated in the clustering process, the driver information condition corresponding to each cluster, the number of clusters, and other various data generated in the portrait process, which are referred to as intermediate values corresponding to the granularity in this embodiment, are stored in the database after each portrait calculation, so that subsequent portrait updating can be conveniently performed.
According to the method, a single logistics driver image is obtained through big data processing, on the basis, the similarity of each quantitative label in the logistics driver image is calculated, and then the group image of the logistics driver is obtained through a clustering method. For a logistics service provider with certain manpower and vehicle resources, the logistics drivers are classified into a plurality of logistics driver image classes with small difference in class and large difference between classes through the method, the structural condition of the manpower transportation capacity service capability of the logistics service provider can be reflected from multiple directions, so that customers can know more about the platform, and data support is provided for a manager to improve the platform service subsequently. The logistics driver image is multi-level clustered, so that the logistics driver image is more comprehensively depicted, the success rate of follow-up accurate recommendation is improved, and the service quality and the competitiveness of the platform are improved. In addition, the group images of the logistics drivers enable the logistics drivers to be clustered according to the images, and offline communication between the drivers of the same group images is facilitated. In addition, the invention considers the calculation of time-route similarity, and incorporates the calculation into the labels required by clustering, and fully utilizes the service dynamic information, so that the group portrait is dynamic and more accords with the actual situation.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Various modifications and alterations to this invention will become apparent to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims (6)

1. A group portrait method for logistic drivers is characterized by comprising the following steps:
s1, qualitative driver image including the granularity and qualitative label;
s2, driver quantitative portrait including quantitative labels for determining driver dimension, vehicle dimension and customer dimension; the driver dimension quantitative label comprises at least a driver order taking time-route label;
s3, converting the qualitative labels into quantitative labels, and calculating the similarity of any two drivers on each quantitative label to obtain the driver portrait similarity; the similarity calculation process of the driver order-receiving time-route comprises the following steps:
a1, the driver order-receiving time-route type matrix is:
Figure FDA0003052365040000011
namely: TDL ═ xij}m×n，1≤i≤m,1≤j≤n；
Where m represents the number of drivers, n route categories, tiDenotes a time period, xijIs shown over a time period tiThe order route of the inner driver i for taking orders is j times;
a2, obtaining the similarity according to the driver order-receiving time-route type matrix:
Figure FDA0003052365040000012
wherein L is a route type, L is a set of route types,
Figure FDA0003052365040000013
represents tiThe number of times that driver d takes an order of type l in the time slot is equivalent to x in the TDL matrixij；
A3, calculating time period tiAnd time period tjThe interval therebetween:
Figure FDA0003052365040000014
a4, calculating the initial time similarity:
Figure FDA0003052365040000015
a5, smoothing the time-route matrix of driver order receiving if driver diAt tkOrders of type l for the route have not been accepted during the time period, i.e.
Figure FDA0003052365040000016
If 0, then use driver diThe number of times of order with type of single route as l in other time periods and tkSimilarity pair of
Figure FDA0003052365040000021
Carrying out prediction; the calculation is as follows:
Figure FDA0003052365040000022
a6, calculating the similarity of the driver order taking time-route according to the data after the smoothing treatment as follows:
Figure FDA0003052365040000023
s4, clustering the driver portrait according to the similarity of the driver portrait;
and S5, generating and updating the group driver portrait according to the clustering result of the step S4.
2. The group portrait method of logistic drivers as claimed in claim 1, wherein the step S3 comprises the following steps:
s31, converting the qualitative labels into quantitative labels, normalizing the quantitative labels of continuous value types, and performing binary variation quantization on the quantitative labels of discrete category values;
s32, calculating the similarity of the two driver images processed in the step S31 on a certain quantitative label;
and S33, summing the similarity of all quantitative labels of the two drivers to obtain the similarity between the two drivers.
3. The crowd portrayal method of logistic drivers as claimed in claim 2, wherein step S32 comprises: the quantitative label similarity calculation for a binary variable is:
Figure FDA0003052365040000024
wherein, sim (U)i,Uj) Driver representation UiAnd driver portrait UjSimilarity on this binary variable quantitative Label, label (U)i) Driver representation UiThe value on the quantitative Label, label (U)j) Driver representation UjA value on the quantitative label;
quantitative label similarity calculation for some non-binary variable:
Figure FDA0003052365040000025
wherein, sim (U)i,Uj) Driver representation UiAnd driver portrait UjSimilarity on the non-binary quantitative Label, labelk(Ui) Driver representation UiValue on the kth quantitative Label, labelk(Uj) Driver representation UjThe value on the kth quantitative label.
4. The crowd portrayal method of logistic drivers as claimed in claim 1, wherein the driver dimension label of step S2 comprises: cleaning basic attributes of a driver in a logistics driver database owned by a logistics service provider, wherein the basic attributes comprise name, age, gender and contact information;
the driver behavior information including order receiving times, freight times and freight mileage is excavated by jointly checking the driver data sheet and the order data sheet;
and (4) jointly checking a driver data table, an order table, a client table and a cargo table, and mining the interest preference of the driver, including the order receiving time preference, the cargo route preference, the transaction client preference and the order receiving price preference.
5. The crowd portrayal method of logistic drivers as claimed in claim 4, wherein the vehicle dimension label of step S2 comprises: cleaning vehicle attributes including length, width, height and load limit in a logistics vehicle database owned by a logistics service provider; and acquiring the real-time position of the vehicle through a third-party real-time positioning service module interface.
6. The crowd portrayal method of logistic drivers as claimed in claim 4, wherein the step S2 comprises the step of: obtaining customer characteristics including names, industries and contact information in a historical transaction database of a driver of the logistics vehicle;
checking historical transaction data and an order data table of a driver to obtain customer behaviors including order times and cash settlement times;
the driver data table, the goods data table, and the order data table are co-reviewed to obtain customer interest preferences including order time preferences, goods characteristics preferences, driver trading preferences, order price preferences, and order route preferences.
CN201910376458.XA 2019-05-07 2019-05-07 Group portrait method for logistic drivers Active CN110097287B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910376458.XA CN110097287B (en) 2019-05-07 2019-05-07 Group portrait method for logistic drivers

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910376458.XA CN110097287B (en) 2019-05-07 2019-05-07 Group portrait method for logistic drivers

Publications (2)

Publication Number Publication Date
CN110097287A CN110097287A (en) 2019-08-06
CN110097287B true CN110097287B (en) 2021-06-22

Family

ID=67446979

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910376458.XA Active CN110097287B (en) 2019-05-07 2019-05-07 Group portrait method for logistic drivers

Country Status (1)

Country Link
CN (1) CN110097287B (en)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111126909B (en) * 2019-12-20 2023-08-08 贵阳货车帮科技有限公司 Data processing method, device, equipment and storage medium for goods source route
CN111125545A (en) * 2019-12-24 2020-05-08 北京旷视科技有限公司 Target object determination method and device and electronic equipment
CN111369313A (en) * 2020-02-03 2020-07-03 深圳市道旅旅游科技股份有限公司 Processing method and device for house-ordering failure order, computer equipment and storage medium
CN112613813A (en) * 2020-12-22 2021-04-06 天津五八到家货运服务有限公司 Method, equipment and medium for pushing freight work scheme
CN112819232A (en) * 2021-02-04 2021-05-18 北京建筑大学 People flow characteristic prediction method and device based on card punching data

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108108451A (en) * 2017-12-27 2018-06-01 合肥美的智能科技有限公司 The group of subscribers portrait acquisition methods and device of group
CN108537586A (en) * 2018-03-30 2018-09-14 杭州米趣网络科技有限公司 Data processing method and device based on user's portrait
CN108985819A (en) * 2018-06-15 2018-12-11 天津五八到家科技有限公司 Driver's portrait method, system and equipment
CN109033332A (en) * 2018-07-20 2018-12-18 汉纳森（厦门）数据股份有限公司 Driving behavior analysis method, medium and system
CN109635852A (en) * 2018-11-26 2019-04-16 汉纳森（厦门）数据股份有限公司 A kind of building of user's portrait and clustering method based on multidimensional property
CN109636235A (en) * 2018-12-26 2019-04-16 北京汽车研究总院有限公司 The determination method and processing system of driving behavior portrait model

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9418491B2 (en) * 2014-09-22 2016-08-16 Brian K. Phillips Method and system for automatically identifying a driver by creating a unique driver profile for a vehicle from driving habits
CN106373331A (en) * 2016-09-28 2017-02-01 百度在线网络技术（北京）有限公司 Riding early-warning method and device
CN109117869B (en) * 2018-07-20 2021-09-17 汉纳森（厦门）数据股份有限公司 User portrayal method, medium and system

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108108451A (en) * 2017-12-27 2018-06-01 合肥美的智能科技有限公司 The group of subscribers portrait acquisition methods and device of group
CN108537586A (en) * 2018-03-30 2018-09-14 杭州米趣网络科技有限公司 Data processing method and device based on user's portrait
CN108985819A (en) * 2018-06-15 2018-12-11 天津五八到家科技有限公司 Driver's portrait method, system and equipment
CN109033332A (en) * 2018-07-20 2018-12-18 汉纳森（厦门）数据股份有限公司 Driving behavior analysis method, medium and system
CN109635852A (en) * 2018-11-26 2019-04-16 汉纳森（厦门）数据股份有限公司 A kind of building of user's portrait and clustering method based on multidimensional property
CN109636235A (en) * 2018-12-26 2019-04-16 北京汽车研究总院有限公司 The determination method and processing system of driving behavior portrait model

Also Published As

Publication number Publication date
CN110097287A (en) 2019-08-06

Similar Documents

Publication Publication Date Title
CN110097287B (en) Group portrait method for logistic drivers
CN110392899B (en) Dynamic feature selection for model generation
US20230073709A1 (en) Image analysis and identification using machine learning with output estimation
US10936947B1 (en) Recurrent neural network-based artificial intelligence system for time series predictions
US20210357835A1 (en) Resource Deployment Predictions Using Machine Learning
US9916584B2 (en) Method and system for automatic assignment of sales opportunities to human agents
US10748195B2 (en) Database image matching using machine learning with output estimation
CN113553540A (en) Commodity sales prediction method
US11543250B2 (en) Securitized and encrypted data for vehicle service scheduling and dispatch devices (SSDD) and systems that provide improved operations and outcomes
US11182841B2 (en) Prospect recommendation
CN111915366A (en) User portrait construction method and device, computer equipment and storage medium
US20080255900A1 (en) Process for identifying potential customers for business outsourcing
CN112068962B (en) Cloud rendering resource exchange method based on deep learning
CN114663198A (en) Product recommendation method, device and equipment based on user portrait and storage medium
CN113424207B (en) System and method for efficiently training understandable models
US12019596B2 (en) System and method for enriching and normalizing data
CN113516496A (en) Advertisement conversion rate pre-estimation model construction method, device, equipment and medium thereof
CN114997916A (en) Prediction method, system, electronic device and storage medium of potential user
CN111738812B (en) Information pushing method and system based on user group micro-segmentation
CN112132396A (en) Customer relationship distribution method and system based on intelligent matching
CN111401935B (en) Resource allocation method, device and storage medium
Brackmann et al. Identifying Application Areas for Machine Learning in the Retail Sector: A Literature Review and Interview Study
Inyang et al. Unsupervised learning framework for customer requisition and behavioral pattern classification
Verroios et al. Client clustering for hiring modeling in work marketplaces
Chen et al. Strategic Decision-making Processes of NPD by Hybrid Classification Model Techniques

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