CN110517059A - A kind of fashion handbag sales forecasting method based on random forest - Google Patents
A kind of fashion handbag sales forecasting method based on random forest Download PDFInfo
- Publication number
- CN110517059A CN110517059A CN201910611386.2A CN201910611386A CN110517059A CN 110517059 A CN110517059 A CN 110517059A CN 201910611386 A CN201910611386 A CN 201910611386A CN 110517059 A CN110517059 A CN 110517059A
- Authority
- CN
- China
- Prior art keywords
- feature
- random forest
- day
- sales
- divided
- 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.)
- Pending
Links
Classifications
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
Landscapes
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of fashion handbag sales forecasting method based on random forest, includes the following steps: S1, according to the historic sales data that city obtains, divides by week, fills in missing values with data center's measurement；The feature of model is established in S2, analysis, carries out Feature Engineering；The feature specifically: (1) according to actual sales volume situation, to city divided rank；(2) consider that living standard influences, the urban economy speedup in season where obtaining；S3 establishes Random Forest model；The present invention is in addition to considering historic sales data, it is also contemplated that other external influence factors are classified as feature；Using Random Forest model, Random Forest model, which has, does not have to beta pruning, does not have to carry out Feature Selection, computing cost is small, the fast advantage of training speed；This method difficulty is moderate, therefore is easy to implement, practical.
Description
Technical field
The present invention relates to product marketing forecast technical fields, and in particular to a kind of fashion handbag sale based on random forest
Prediction technique.
Background technique
With the development of the social economy, people's level of consumption is also gradually increased, brand fashion women's bag has been that people are common
The consumer goods, for brand side, the sales volume in a period of time is directly related to the achievement of company, if it is possible to know one
The potential sale of section time product can be the short-term planning of production line, stress to mention for the product design and publicity of brand side
For reference.So the sales forecast for fashion handbag is of great significance.
Fast different from common product update iteration, brand fashion handbag has certain reserve value, some productions due to valence height
Product become classics, it is evident that have long life cycle.For the sales volume prediction of the product of long life cycle, need to consider longer
Temporal characteristics.Moreover, if in scheme only use divide the period total sales volume predicted as feature, this for
The sales volume prediction of independent product does not have good effect.Finally, considering external factor, such as some products also can be by weather
It influencing, fashion handbag is often made using corium, is easy crumple in a humid environment, and it is mouldy, also influence whether consumer's
Utilization rate, so the case where also comprehensively considering weather, while in different areas, sales volume is also different.
For the sales volume prediction of product, in the prior art, domestic Wang Rui discloses one kind " based on Gong Baici curve
Durable goods sales forecast naive model " utilizes Gong Baici mathematic curve model and sales data over the years, obtains undetermined parameter,
Finally obtain simple Demand Forecast Model；Chen Yinguang et al. discloses a kind of " based on the clothes for improving the method for weighted moving average
Sales forecast ", which introduce the concepts of trend and the method for solving prediction lag phenomenon, and it is mobile flat to propose improved weighting
Equal method, is predicted in conjunction with historic sales data.But the model of these methods is all relatively simple, and only considered history
Influence characteristic of sales data, the characteristics of not accounting for product itself and external environment etc. is therefore obtained pre- as feature
Survey result accuracy is not high, and prediction effect still has to be hoisted.
Summary of the invention
The purpose of the present invention is to overcome the shortcomings of the existing technology and deficiency, provides a kind of fashion hand based on random forest
Bag sales forecasting method, this method solve only using historical sales, do not account for independent product feature and external environment it is special
Sign, the poor problem of independent product prediction effect, can comprehensively consider the feature of product, historical sales, external environment spy
Sign achievees the effect that improve precision of prediction using random forests algorithm.
The purpose of the invention is achieved by the following technical solution:
A kind of fashion handbag sales forecasting method based on random forest, includes the following steps:
S1 is divided according to the historic sales data that city obtains by week, fills in missing values with data center's measurement；
The feature of model is established in S2, analysis, carries out Feature Engineering；The feature specifically:
(1) according to actual sales volume situation, to city divided rank；
(2) consider that living standard influences, the urban economy speedup in season where obtaining；
(3) this week weather characteristics obtain its overall weather condition, are divided into fine day, rainy day, cloudy day, this is nominal type feature,
It is encoded into three-dimensional binary feature, i.e., is set as 1 in the position of corresponding states value, remaining position is both configured to 0；
(4) whether day is promoted, this week, there are important promotion day such as 6.18, double 11 to divide class 4 into, and New Year's Day, the Spring Festival divide into
Grade 3, the Dragon Boat Festival, National Day divide grade 2 into, and remaining festivals or holidays divide grade 1, common day mark 0 into；
(5) fashion handbag product style divides: mainly there are both shoulders packet, satchel, handbag, hand to take and wraps, nominal type feature, together
Sample is encoded into binary feature；
(6) whether there is or not shoulder belt, be binary feature whether there is or not accessories, whether there is or not waterproof coating, " having " then marks 1, and "None" then marks 0；
(7) when season prevalence color be divided into three kinds: cool colour, warm colour, intermediate colour system；This is nominal type feature, by it
It is encoded into three-dimensional binary feature.
S3 establishes Random Forest model；
Data set is divided into training set S and test set T, feature quantity M by S3.1；
S3.2 puts back to ground sample drawn from initial data concentration, and sample size and training set S are in the same size, Yi Gongjin
Row k times, forms k decision tree, this k training set starts to train as the root node of corresponding single decision tree；
S3.3 randomly selects m feature from M feature, meets m < < M, using Minimum Mean Square Error as division principle, utilizes
Feature where this m feature finds optimal partition point, is set as leaf node, output valve for present node if meeting termination condition
For the average value of the sample set；
S3.4 is repeated k times, generates k decision tree, forms Random Forest model；
S3.5 treats forecast set T and is predicted using the obtained Random Forest model of S3.4, exports predicted value.
The present invention have compared with prior art it is below the utility model has the advantages that
The present invention is in addition to considering historic sales data, it is also contemplated that other external influence factors are classified as feature；Using with
Machine forest model, Random Forest model, which has, does not have to beta pruning, does not have to carry out Feature Selection, and computing cost is small, and training speed is fast
Advantage；This method difficulty is moderate, therefore is easy to implement, practical.
Detailed description of the invention
Fig. 1 is flow diagram of the invention.
Specific embodiment
Present invention will now be described in further detail with reference to the embodiments and the accompanying drawings, but embodiments of the present invention are unlimited
In this.
As shown in Figure 1, a kind of fashion handbag sales forecasting method based on random forest, includes the following steps:
S1 is divided according to the historic sales data that city obtains by week, fills in missing values with data center's measurement；
The feature of model is established in S2, analysis, carries out Feature Engineering；The feature specifically:
(1) according to actual sales volume situation, to city divided rank；
(2) consider that living standard influences, the urban economy speedup in season where obtaining；
(3) this week weather characteristics obtain its overall weather condition, are divided into fine day, rainy day, cloudy day, this is nominal type feature,
It is encoded into three-dimensional binary feature, i.e., is set as 1 in the position of corresponding states value, remaining position is both configured to 0；
(4) whether day is promoted, this week, there are important promotion day such as 6.18, double 11 to divide class 4 into, and New Year's Day, the Spring Festival divide into
Grade 3, the Dragon Boat Festival, National Day divide grade 2 into, and remaining festivals or holidays divide grade 1, common day mark 0 into；
(5) fashion handbag product style divides: mainly there are both shoulders packet, satchel, handbag, hand to take and wraps, nominal type feature, together
Sample is encoded into binary feature；
(6) whether there is or not shoulder belt, be binary feature whether there is or not accessories, whether there is or not waterproof coating, " having " then marks 1, and "None" then marks 0；
(7) when season prevalence color be divided into three kinds: cool colour, warm colour, intermediate colour system；This is nominal type feature, by it
It is encoded into three-dimensional binary feature.
S3 establishes Random Forest model；
Data set is divided into training set S and test set T, feature quantity M by S3.1；
S3.2 puts back to ground sample drawn from initial data concentration, and sample size and training set S are in the same size, Yi Gongjin
Row k times, forms k decision tree, this k training set starts to train as the root node of corresponding single decision tree；
S3.3 randomly selects m feature from M feature, meets m < < M, using Minimum Mean Square Error as division principle, utilizes
Feature where this m feature finds optimal partition point, is set as leaf node, output valve for present node if meeting termination condition
For the average value of the sample set；
S3.4 is repeated k times, generates k decision tree, forms Random Forest model；
S3.5 treats forecast set T and is predicted using the obtained Random Forest model of S3.4, exports predicted value.
The invention reside in for the brand fashion handbag product market with long life cycle, random forest prediction is proposed
Method, and the shortcomings that overcome existing sales forecast, solve only using historical sales, do not account for independent product feature and
The poor problem of external environment feature, independent product prediction effect；This method will comprehensively consider feature, the historical sales of product
Amount, the feature of external environment achieve the effect that improve precision of prediction using random forests algorithm.
The present invention is in addition to considering historic sales data, it is also contemplated that other external influence factors are classified as feature；Using with
Machine forest model, Random Forest model, which has, does not have to beta pruning, does not have to carry out Feature Selection, and computing cost is small, and training speed is fast
Advantage；This method difficulty is moderate, therefore is easy to implement, practical.
Above-mentioned is the preferable embodiment of the present invention, but embodiments of the present invention are not limited by the foregoing content,
His any changes, modifications, substitutions, combinations, simplifications made without departing from the spirit and principles of the present invention, should be
The substitute mode of effect, is included within the scope of the present invention.
Claims (1)
1. a kind of fashion handbag sales forecasting method based on random forest, which is characterized in that include the following steps:
S1 is divided according to the historic sales data that city obtains by week, fills in missing values with data center's measurement；
The feature of model is established in S2, analysis, carries out Feature Engineering；The feature specifically:
(1) according to actual sales volume situation, to city divided rank；
(2) consider that living standard influences, the urban economy speedup in season where obtaining；
(3) this week weather characteristics obtain its overall weather condition, are divided into fine day, rainy day, cloudy day, this is nominal type feature, by it
It is encoded into three-dimensional binary feature, i.e., is set as 1 in the position of corresponding states value, remaining position is both configured to 0；
(4) whether day is promoted, this week, there are important promotion day such as 6.18, double 11 to divide class 4 into, and New Year's Day, the Spring Festival divide grade into
3, the Dragon Boat Festival, National Day divide grade 2 into, and remaining festivals or holidays divide grade 1, common day mark 0 into；
(5) fashion handbag product style divides: mainly having both shoulders packet, satchel, handbag, hand to take packet, nominal type feature is same to compile
Code is at binary feature；
(6) whether there is or not shoulder belt, be binary feature whether there is or not accessories, whether there is or not waterproof coating, " having " then marks 1, and "None" then marks 0；
(7) when season prevalence color be divided into three kinds: cool colour, warm colour, intermediate colour system；This is nominal type feature, is encoded
At three-dimensional binary feature.
S3 establishes Random Forest model；
Data set is divided into training set S and test set T, feature quantity M by S3.1；
S3.2 puts back to ground sample drawn from initial data concentration, and sample size and training set S are in the same size, carry out k altogether
It is secondary, k decision tree is formed, this k training set starts to train as the root node of corresponding single decision tree；
S3.3 randomly selects m feature from M feature, meets m < < M, using Minimum Mean Square Error as division principle, utilizes this m
Feature finds feature where optimal partition point, present node is set as leaf node if meeting termination condition, output valve is should
The average value of sample set；
S3.4 is repeated k times, generates k decision tree, forms Random Forest model；
S3.5 treats forecast set T and is predicted using the obtained Random Forest model of S3.4, exports predicted value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910611386.2A CN110517059A (en) | 2019-07-08 | 2019-07-08 | A kind of fashion handbag sales forecasting method based on random forest |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910611386.2A CN110517059A (en) | 2019-07-08 | 2019-07-08 | A kind of fashion handbag sales forecasting method based on random forest |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110517059A true CN110517059A (en) | 2019-11-29 |
Family
ID=68623782
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910611386.2A Pending CN110517059A (en) | 2019-07-08 | 2019-07-08 | A kind of fashion handbag sales forecasting method based on random forest |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110517059A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112001757A (en) * | 2020-08-26 | 2020-11-27 | 中山世达模型制造有限公司 | Sales order prediction method |
| CN112668802A (en) * | 2021-01-05 | 2021-04-16 | 广东工业大学 | Construction carbon emission prediction method based on design parameters |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103617459A (en) * | 2013-12-06 | 2014-03-05 | 李敬泉 | Commodity demand information prediction method under multiple influence factors |
| CN107180362A (en) * | 2017-05-03 | 2017-09-19 | 浙江工商大学 | Retail commodity sales forecasting method based on deep learning |
| CN109034255A (en) * | 2018-08-02 | 2018-12-18 | 深圳码隆科技有限公司 | The prediction technique and device of clothes sales volume |
| CN109345011A (en) * | 2018-09-19 | 2019-02-15 | 中冶赛迪重庆信息技术有限公司 | A kind of Air-conditioning Load Prediction method and system returning forest based on depth |
| CN109360022A (en) * | 2018-10-15 | 2019-02-19 | 广东工业大学 | A kind of market Sales Volume of Commodity prediction technique, device and equipment based on data mining |
-
2019
- 2019-07-08 CN CN201910611386.2A patent/CN110517059A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103617459A (en) * | 2013-12-06 | 2014-03-05 | 李敬泉 | Commodity demand information prediction method under multiple influence factors |
| CN107180362A (en) * | 2017-05-03 | 2017-09-19 | 浙江工商大学 | Retail commodity sales forecasting method based on deep learning |
| CN109034255A (en) * | 2018-08-02 | 2018-12-18 | 深圳码隆科技有限公司 | The prediction technique and device of clothes sales volume |
| CN109345011A (en) * | 2018-09-19 | 2019-02-15 | 中冶赛迪重庆信息技术有限公司 | A kind of Air-conditioning Load Prediction method and system returning forest based on depth |
| CN109360022A (en) * | 2018-10-15 | 2019-02-19 | 广东工业大学 | A kind of market Sales Volume of Commodity prediction technique, device and equipment based on data mining |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112001757A (en) * | 2020-08-26 | 2020-11-27 | 中山世达模型制造有限公司 | Sales order prediction method |
| CN112668802A (en) * | 2021-01-05 | 2021-04-16 | 广东工业大学 | Construction carbon emission prediction method based on design parameters |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Parikh et al. | Impact of rural industrialization on village life and economy: a social accounting matrix approach | |
| CN110517059A (en) | A kind of fashion handbag sales forecasting method based on random forest | |
| CN107909294A (en) | A kind of electric company's customer value evaluation method based on customer portrait technology | |
| CN106600959A (en) | Traffic congestion index-based prediction method | |
| CN110147919A (en) | A kind of public bicycles automatic scheduling method based on price competition mechanism | |
| Thirlwall et al. | A model of regional growth rate differences on Kaldorian lines | |
| CN104134096A (en) | Intelligent load prediction method based on province and prefecture integration | |
| CN105761489A (en) | Cubic exponential smoothing optimal method of traffic flow prediction | |
| Han et al. | Free shipping policy for imported cross-border e-commerce platforms | |
| Wu et al. | Application of neural network model based on multispecies evolutionary genetic algorithm to planning and design of diverse plant landscape | |
| Lutz | The Essentials of Capital Theory | |
| Mottura et al. | Agriculture and agribusiness: transformations and trends in Italy | |
| Choy et al. | Staple trade, real wages, and living standards in Singapore, 1870–1939 | |
| CN109685197A (en) | A kind of optimization method based on the power supply set meal for improving particle swarm algorithm | |
| Belomestnov et al. | Quality of investment processes in the regions | |
| Kuznetsova et al. | Evaluation of the competitive potential of the economic development of the country: theoretical aspects and Russian practice | |
| Zhang et al. | Virtual water flows and drivers in the international trade of agricultural products of the regional comprehensive economic partnership | |
| Kazak | Analysis of dynamics of demand, revenue and ergonomic aspects of tourism | |
| Rowens | Luxury Consumption | |
| Fan | Economic development evaluation model based on entropy AHP | |
| Habuki | THE UNITED STATES HOP INDUSTRY: STRUCTURAL ANALYSIS AND FORECASTS. | |
| Alshamry | Developing a Strategy for Increasing the Efficiency of the Marketing of Selected Agricultural Produce in the United Arab Emirates. | |
| Krantz et al. | 2 How it all began | |
| Kaser | Reform in the USSR and China | |
| Krantz et al. | How it all began: On structural periods |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191129 |