CN109447716A - Method for Sales Forecast method and server based on Recognition with Recurrent Neural Network - Google Patents
Method for Sales Forecast method and server based on Recognition with Recurrent Neural Network Download PDFInfo
- Publication number
- CN109447716A CN109447716A CN201811334201.XA CN201811334201A CN109447716A CN 109447716 A CN109447716 A CN 109447716A CN 201811334201 A CN201811334201 A CN 201811334201A CN 109447716 A CN109447716 A CN 109447716A
- Authority
- CN
- China
- Prior art keywords
- recognition
- neural network
- sales volume
- recurrent neural
- volume data
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Abstract
The present invention relates to technical field of data processing, and the present invention is to solve the not high problems of the accuracy of existing Method for Sales Forecast, the Method for Sales Forecast method based on Recognition with Recurrent Neural Network, mainly comprise the steps that the multiple groups history sales volume data for obtaining product to be predicted；The multiple groups history sales volume data are split as training dataset and validation data set；Using the training dataset as training sample, training obtains Recognition with Recurrent Neural Network model, carries out verifying according to the validation data set until the Recognition with Recurrent Neural Network model meets Method for Sales Forecast requirement；According to the Recognition with Recurrent Neural Network model, sales volume data of the product to be predicted in the designated time period after current time are determined.The accuracy of product Method for Sales Forecast can be improved in method of the invention, the Method for Sales Forecast suitable for various product.
Description
Technical field
The present invention relates to technical field of data processing, a kind of Method for Sales Forecast method and server are related in particular to.
Background technique
In merchandise sales industry, the prediction of sales volume is an indispensable important indicator, and Method for Sales Forecast is usually basis
The history sales volume of product predicts that the sales volume situation in product future, the prior art generally use regression analysis realization,
Principle is to fit a function to series of discrete point, small as far as possible with true value difference by the calculated value of function.But this
Kind method has a supposed premise, i.e. data are obeyed certain and were both distributed, and parameters are further solved under the premise of this.But
It is for Method for Sales Forecast, sales volume data distribution in practice is different, and often generalization is poor for conventional regression analysis, passes through biography
The regression analysis of system is not high come the prediction accuracy for carrying out sales volume.
Summary of the invention
The invention aims to solve the problems, such as that the accuracy of existing Method for Sales Forecast is not high, propose a kind of based on circulation mind
Method for Sales Forecast method and server through network.
The technical proposal adopted by the invention to solve the above technical problems is that: the Method for Sales Forecast side based on Recognition with Recurrent Neural Network
Method, which comprises the following steps:
S1. the multiple groups history sales volume data of product to be predicted are obtained, the history sales volume data include historical time sequence
And corresponding true sales volume data；
S2. the multiple groups history sales volume data are split as training dataset and validation data set；
S3. using the training dataset as training sample, training obtains Recognition with Recurrent Neural Network model, by the verifying number
According to the historical time sequence inputting of collection to the Recognition with Recurrent Neural Network model, the corresponding prediction pin of the historical time sequence is obtained
Measure data；
S4. calculate the error of the prediction sales volume data relative to corresponding true sales volume data, and by the error with
Preset threshold is compared, if error is greater than preset threshold, by this group of history pin of the corresponding verify data concentration of the error
It measures data and training dataset is added, enter step S3, otherwise, enter step S5；
S5. according to the Recognition with Recurrent Neural Network model, determine the product to be predicted in specified after current time
Between sales volume data in section.
It further, is the accuracy for improving Method for Sales Forecast, after the history sales volume data for obtaining product to be predicted
Further include:
The history sales volume data are pre-processed.
Specifically, in order to solve the problems, such as that the range of each sales volume data is different, it is described that the history sales volume data are carried out
Pretreatment includes:
The history sales volume data are normalized:
In formula, x 'iFor the corresponding true sales volume data of i-th of time series after normalized, xiFor i-th of time
The corresponding true sales volume data of sequence, u are the mean value of true sales volume data, and σ is variance；
The historical time sequence inputting by the validation data set obtains described to the Recognition with Recurrent Neural Network model
The corresponding prediction sales volume data of historical time sequence include:
By the historical time sequence inputting of the validation data set to the Recognition with Recurrent Neural Network model, output valve is obtained,
The corresponding prediction sales volume data of i-th of time series are calculated according to the output valve.
Specifically, to further increase the accuracy of Method for Sales Forecast, it is described using the training dataset as training sample
Include:
Choose for training t group training data, every group of training data respectively include historical time sequence and it is corresponding very
Real sales volume data；
In the t group training data, since first group of training data, first group is chosen to n-th group training data work
For input, (n+1)th group of training data is chosen as output, is configured to one group of training sample；
In the t group training data, since second group of training data, second group to (n+1)th group training data is chosen
As input, the n-th+2 group training data is chosen as output, is configured to another group of training sample.
Specifically, to further increase the accuracy of Method for Sales Forecast, the training obtains Recognition with Recurrent Neural Network model and includes:
Set Recognition with Recurrent Neural Network hidden layer number and input layer number；
It is trained according to training sample, the mean square error of Recognition with Recurrent Neural Network is minimized using gradient descent method；
It obtains Recognition with Recurrent Neural Network model and saves.
For the accuracy for further increasing Method for Sales Forecast, the Method for Sales Forecast method based on Recognition with Recurrent Neural Network is also wrapped
It includes: updating the Recognition with Recurrent Neural Network model.
Further, the sales volume of the following specified time and given client are predicted to realize, the history sales volume number
According to further including historical time sequence and the corresponding true sales volume data of client；
The step S3 further include: the historical time sequence of the validation data set and client are input to the circulation mind
Through network model, the historical time sequence and the corresponding prediction sales volume data of client are obtained；
The step S5 further include: according to the Recognition with Recurrent Neural Network model, determine the product to be predicted when current
The sales volume data of designated time period and given client after quarter.
Based on the above-mentioned technical proposal, the present invention also provides a kind of server, the server include processor, memory and
Communication bus；
The communication bus is for realizing the connection communication between processor and memory；
The processor is used to execute one or more program in memory, above-mentioned based on circulation nerve to realize
The step of Method for Sales Forecast method of network.
The beneficial effects of the present invention are: the Method for Sales Forecast method and service of the present invention based on Recognition with Recurrent Neural Network
Device, by establishing Recognition with Recurrent Neural Network model, according to the training dataset in history sales volume data to Recognition with Recurrent Neural Network model
It is trained, the accuracy of recirculating network neural model is verified by the validation data set in history sales volume data, is followed
After ring neural network model is verified, sales volume of the product to be predicted in the designated time period after current time is carried out pre-
It surveys, to improve the accuracy of Method for Sales Forecast.
Detailed description of the invention
Fig. 1 is the flow diagram of the Method for Sales Forecast method based on Recognition with Recurrent Neural Network described in the embodiment of the present invention.
Specific embodiment
Embodiments of the present invention are described in detail below in conjunction with attached drawing.
Method for Sales Forecast method of the present invention based on Recognition with Recurrent Neural Network, comprising the following steps: S1. obtains to be predicted
The multiple groups history sales volume data of product, the history sales volume data include historical time sequence and corresponding true sales volume data；
S2. the multiple groups history sales volume data are split as training dataset and validation data set；S3. using the training dataset as
Training sample, training obtain Recognition with Recurrent Neural Network model, the historical time sequence inputting of the validation data set are followed to described
Ring neural network model obtains the corresponding prediction sales volume data of the historical time sequence；S4. the prediction sales volume data are calculated
It is compared relative to the error of corresponding true sales volume data, and by the error with preset threshold, is preset if error is greater than
This group of history sales volume data that the corresponding verify data of the error is concentrated are added training dataset, enter step S3 by threshold value,
Otherwise, S5 is entered step；S5. according to the Recognition with Recurrent Neural Network model, determine the product to be predicted after current time
Designated time period in sales volume data.
No need to reserve adopted type function for Recognition with Recurrent Neural Network, can theoretically be approached by multiple nonlinear transformation any
Function, simultaneously for the corresponding sales volume data of time series, current sales volume data often with history sales volume data strong correlation, are recycled
Neural network can remember previous state, with the later state of previous status predication, specifically be exactly, according to history pin
The training that the training dataset in data carries out Recognition with Recurrent Neural Network model is measured, according to the validation data set in history sales volume data
The accuracy of recirculating network neural model is verified, after Recognition with Recurrent Neural Network model is verified, is existed to product to be predicted
The sales volume in designated time period after current time is predicted.
Embodiment
Method for Sales Forecast method described in the embodiment of the present invention based on Recognition with Recurrent Neural Network, as shown in Figure 1, including following step
It is rapid:
S1. the multiple groups history sales volume data of product to be predicted are obtained, the history sales volume data include historical time sequence
And corresponding true sales volume data；
Wherein, product to be predicted be pending Method for Sales Forecast product, the prediction product can be set as needed or
Selection, historical time sequence are designated time period sequence, such as in January, 2017 to December, the 1st week to the 8th week, 2010 to 2017
Year, designated time period can according to need setting.
S2. the multiple groups history sales volume data are split as training dataset and validation data set；
Fractionation is split to the quantity of multiple groups history sales volume data, i.e., using a part of history sales volume data as instruction
Practice data set, another part is used as validation data set, training dataset and verify data concentrate including historical time sequence and
Corresponding sales volume data.
S3. using the training dataset as training sample, training obtains Recognition with Recurrent Neural Network model, by the verifying number
According to the historical time sequence inputting of collection to the Recognition with Recurrent Neural Network model, the corresponding prediction pin of the historical time sequence is obtained
Measure data；
Wherein, training dataset is for training Recognition with Recurrent Neural Network model, specifically, can first preset circulation mind
Through network rudimentary model, initial parameter is set in the Recognition with Recurrent Neural Network rudimentary model, training dataset is substituted into the circulation
It in neural network rudimentary model, is constantly trained, and then adjusts the parameter of the Recognition with Recurrent Neural Network rudimentary model, thus
To Recognition with Recurrent Neural Network model；Validation data set passes through verifying for verifying to the Recognition with Recurrent Neural Network model trained
Data set carries out the verifying of the Recognition with Recurrent Neural Network model, specifically, the historical time sequence inputting of validation data set extremely should
Recognition with Recurrent Neural Network model obtains the corresponding prediction sales volume data of the historical time sequence, and then judges the circulation nerve net
Whether network model meets the requirement of Method for Sales Forecast；Validation data set and training dataset are separated, circulation nerve net can be improved
The authenticity of network model verifying, to improve the accuracy that Recognition with Recurrent Neural Network model carries out Method for Sales Forecast.
S4. calculate the error of the prediction sales volume data relative to corresponding true sales volume data, and by the error with
Preset threshold is compared, if error is greater than preset threshold, by this group of history pin of the corresponding verify data concentration of the error
It measures data and training dataset is added, enter step S3, otherwise, enter step S5；
It is understood that the prediction sales volume data and verify data that are calculated according to Recognition with Recurrent Neural Network model are concentrated
True sales volume data compare, if error between the two is greater than default error threshold, then it represents that the Recognition with Recurrent Neural Network
Model is unsatisfactory for the requirement of Method for Sales Forecast, this group of history sales volume data for concentrating the corresponding verify data of the error is needed to add
Enter training dataset, and be trained again, enters step S3, if error between the two is not more than default error threshold,
It indicates that the Recognition with Recurrent Neural Network model meets the requirement of Method for Sales Forecast, can be used for the Method for Sales Forecast of product to be predicted, into step
The threshold value of rapid S5, the error can be set as needed.
S5. according to the Recognition with Recurrent Neural Network model, determine the product to be predicted in specified after current time
Between sales volume data in section.
Wherein, the designated time period after current time is relevant to history sales volume data, and e.g., history sales volume data are
Product to be predicted is in the sales volume data in January to December, then the designated time period after current time is after current time
One month sales volume data.
Optionally, after the history sales volume data for obtaining product to be predicted further include:
The history sales volume data are pre-processed；
Due to being influenced by certain factors, the history sales volume data of product to be predicted there may be in the short time significantly
The case where increasing or reducing, it is therefore desirable to which history sales volume data are pre-processed.
Optionally, it is described to the history sales volume data carry out pretreatment include:
The history sales volume data are normalized:
In formula, x 'iFor the corresponding true sales volume data of i-th of time series after normalized, xiFor i-th of time
The corresponding true sales volume data of sequence, u are the mean value of true sales volume data, and σ is variance；
The historical time sequence inputting by the validation data set obtains described to the Recognition with Recurrent Neural Network model
The corresponding prediction sales volume data of historical time sequence include:
By the historical time sequence inputting of the validation data set to the Recognition with Recurrent Neural Network model, output valve is obtained,
The corresponding prediction sales volume data of i-th of time series are calculated according to the output valve.
It is above-mentioned that the corresponding prediction sales volume data of i-th of time series are calculated according to output valve, specifically, by output valve
The corresponding prediction sales volume data of i-th of time series are obtained along with the mean value u of true sales volume data multiplied by variances sigma；It will go through
After history data are normalized, Recognition with Recurrent Neural Network model stability can be made to restrain.
Optionally, described to include: using the training dataset as training sample
Choose for training t group training data, every group of training data respectively include historical time sequence and it is corresponding very
Real sales volume data；
In the t group training data, since first group of training data, first group is chosen to n-th group training data work
For input, (n+1)th group of training data is chosen as output, is configured to one group of training sample；
In the t group training data, since second group of training data, second group to (n+1)th group training data is chosen
As input, the n-th+2 group training data is chosen as output, is configured to another group of training sample.
History sales volume data are matched, e.g., history sales volume data be in January, 2017 to December sales volume data and
The sales volume data in January, 2018 to October, since in January, 2017,12 groups of sales volume data for choosing in January, 2017 to December are made
For input, the sales volume data in January, 2018 are chosen as output, training sample is configured to and is trained；
Since 2 months 2017,2 months 2017 were chosen to December, 12 groups of sales volume data in January, 2012 are used as input,
The sales volume data for choosing 2 months 2018 are configured to training sample and are trained as output；
And so on, it is that multiple groups training sample is trained to obtain Recognition with Recurrent Neural Network mould by history sales volume data configuration
Type.
Optionally, the training obtains Recognition with Recurrent Neural Network model and includes:
Set Recognition with Recurrent Neural Network hidden layer number and input layer number；
It is trained according to training sample, the mean square error of Recognition with Recurrent Neural Network is minimized using gradient descent method；
It obtains Recognition with Recurrent Neural Network model and saves.
In the training process, mean square error is bigger, indicates that the error of the Recognition with Recurrent Neural Network model is bigger, using under gradient
Drop method solves the minimum value of variance along the direction that gradient declines, thus the Recognition with Recurrent Neural Network for approaching minimal error of recursiveness
Model improves the accuracy of Method for Sales Forecast.
Optionally, the Method for Sales Forecast method based on Recognition with Recurrent Neural Network further include: update the Recognition with Recurrent Neural Network
Model.
With the offset of time, it is possible that the error by the Recognition with Recurrent Neural Network model prediction sales volume increases, lead
Therefore can according to the actual situation the Recognition with Recurrent Neural Network model be updated by causing the accuracy of prediction, e.g., timing is more
Newly.
Optionally, the history sales volume data further include historical time sequence and the corresponding true sales volume data of client；
The step S3 further include: the historical time sequence of the validation data set and client are input to the circulation mind
Through network model, the historical time sequence and the corresponding prediction sales volume data of client are obtained；
The step S5 further include: according to the Recognition with Recurrent Neural Network model, determine the product to be predicted when current
The sales volume data of designated time period and given client after quarter.
It is understood that in order to keep Method for Sales Forecast more targeted, when history sales volume data can also include history
Between sequence and the corresponding true sales volume data of client, by obtain given client history sales volume data, come carry out circulation nerve
The training and verifying of network model, and the Method for Sales Forecast that given client is carried out by the Recognition with Recurrent Neural Network model, Ke Yijin
The accuracy of one step raising Method for Sales Forecast.
Based on the above-mentioned technical proposal, the embodiment of the present invention also provides a kind of server, and the server includes processor, deposits
Reservoir and communication bus；
The communication bus is for realizing the connection communication between processor and memory；
The processor is used to execute one or more program in memory, above-mentioned based on circulation nerve net to realize
The step of Method for Sales Forecast method of network.
Since the above-mentioned Method for Sales Forecast method based on Recognition with Recurrent Neural Network can be realized to product high accuracy to be predicted
Method for Sales Forecast therefore realize that the server of the above-mentioned Method for Sales Forecast method and step based on Recognition with Recurrent Neural Network equally can be real
Now to the Method for Sales Forecast of product high accuracy to be predicted.
Claims (8)
1. the Method for Sales Forecast method based on Recognition with Recurrent Neural Network, which comprises the following steps:
S1. the multiple groups history sales volume data of product to be predicted are obtained, the history sales volume data include historical time sequence and right
The true sales volume data answered；
S2. the multiple groups history sales volume data are split as training dataset and validation data set；
S3. using the training dataset as training sample, training obtains Recognition with Recurrent Neural Network model, by the validation data set
Historical time sequence inputting to the Recognition with Recurrent Neural Network model, obtain the corresponding prediction sales volume number of the historical time sequence
According to；
S4. error of the prediction sales volume data relative to corresponding true sales volume data is calculated, and by the error and is preset
Threshold value is compared, if error is greater than preset threshold, by this group of history sales volume number of the corresponding verify data concentration of the error
According to training dataset is added, S3 is entered step, otherwise, enters step S5；
S5. according to the Recognition with Recurrent Neural Network model, designated time period of the product to be predicted after current time is determined
Interior sales volume data.
2. the Method for Sales Forecast method based on Recognition with Recurrent Neural Network as described in claim 1, which is characterized in that the acquisition is to pre-
It surveys after the history sales volume data of product further include:
The history sales volume data are pre-processed.
3. the Method for Sales Forecast method based on Recognition with Recurrent Neural Network as claimed in claim 2, which is characterized in that described to be gone through to described
History sales volume data carry out pretreatment
The history sales volume data are normalized:
In formula, x 'iFor the corresponding true sales volume data of i-th of time series after normalized, xiFor i-th of time series
Corresponding true sales volume data, u are the mean value of true sales volume data, and σ is variance；
The historical time sequence inputting by the validation data set obtains the history to the Recognition with Recurrent Neural Network model
The corresponding prediction sales volume data of time series include:
By the historical time sequence inputting of the validation data set to the Recognition with Recurrent Neural Network model, output valve is obtained, according to
The corresponding prediction sales volume data of i-th of time series are calculated in the output valve.
4. the Method for Sales Forecast method based on Recognition with Recurrent Neural Network as described in claim 1, which is characterized in that described by the instruction
Practicing data set as training sample includes:
The t group training data for training is chosen, every group of training data respectively includes historical time sequence and corresponding true pin
Measure data；
In the t group training data, since first group of training data, first group is chosen to n-th group training data as defeated
Enter, chooses (n+1)th group of training data as output, be configured to one group of training sample；
In the t group training data, since second group of training data, second group to (n+1)th group training data conduct is chosen
Input chooses the n-th+2 group training data as output, is configured to another group of training sample.
5. the Method for Sales Forecast method based on Recognition with Recurrent Neural Network as described in claim 1, which is characterized in that the training obtains
Recognition with Recurrent Neural Network model includes:
Set Recognition with Recurrent Neural Network hidden layer number and input layer number；
It is trained according to training sample, the mean square error of Recognition with Recurrent Neural Network is minimized using gradient descent method；
It obtains Recognition with Recurrent Neural Network model and saves.
6. the Method for Sales Forecast method based on Recognition with Recurrent Neural Network as described in claim 1, which is characterized in that further include: it updates
The Recognition with Recurrent Neural Network model.
7. the Method for Sales Forecast method based on Recognition with Recurrent Neural Network as described in claim 1, which is characterized in that the history sales volume
Data further include historical time sequence and the corresponding true sales volume data of client；
The step S3 further include: the historical time sequence of the validation data set and client are input to the circulation nerve net
Network model obtains the historical time sequence and the corresponding prediction sales volume data of client；
The step S5 further include: according to the Recognition with Recurrent Neural Network model, determine the product to be predicted current time it
The sales volume data of designated time period and given client afterwards.
8. server, which is characterized in that the server includes processor, memory and communication bus；
The communication bus is for realizing the connection communication between processor and memory；
The processor is used to execute one or more program in memory, to realize such as any one of claim 1 to 7 institute
The step of Method for Sales Forecast method based on Recognition with Recurrent Neural Network stated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811334201.XA CN109447716A (en) | 2018-11-09 | 2018-11-09 | Method for Sales Forecast method and server based on Recognition with Recurrent Neural Network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811334201.XA CN109447716A (en) | 2018-11-09 | 2018-11-09 | Method for Sales Forecast method and server based on Recognition with Recurrent Neural Network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109447716A true CN109447716A (en) | 2019-03-08 |
Family
ID=65550940
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811334201.XA Pending CN109447716A (en) | 2018-11-09 | 2018-11-09 | Method for Sales Forecast method and server based on Recognition with Recurrent Neural Network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109447716A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110458664A (en) * | 2019-08-06 | 2019-11-15 | 上海新共赢信息科技有限公司 | A kind of user's trip information prediction technique, device, equipment and storage medium |
| CN111291657A (en) * | 2020-01-21 | 2020-06-16 | 同济大学 | Crowd counting model training method based on difficult case mining and application |
| CN111652655A (en) * | 2020-06-10 | 2020-09-11 | 创新奇智(上海)科技有限公司 | Commodity sales prediction method and device, electronic equipment and readable storage medium |
| CN113283936A (en) * | 2021-05-28 | 2021-08-20 | 深圳千岸科技股份有限公司 | Sales forecasting method, sales forecasting device and electronic equipment |
| CN113763186A (en) * | 2021-10-22 | 2021-12-07 | 平安科技（深圳）有限公司 | User transfer insurance prediction method, device and equipment based on recurrent neural network |
| CN113837783A (en) * | 2020-06-24 | 2021-12-24 | 上海顺如丰来技术有限公司 | Time series model parameter optimization method and device and computer equipment |
| CN114581157A (en) * | 2022-04-28 | 2022-06-03 | 湖南康道医药有限公司 | Sales prediction method and apparatus based on big data, electronic device, and medium |
| CN115829630A (en) * | 2022-12-16 | 2023-03-21 | 广州飞狮数字科技有限公司 | Method and device for determining sales ratio of shop |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105678588A (en) * | 2016-01-13 | 2016-06-15 | 焦点科技股份有限公司 | Product sales volume fluctuation assessment method based on e-commerce platform |
| CN107239532A (en) * | 2017-05-31 | 2017-10-10 | 北京京东尚科信息技术有限公司 | Data digging method and device |
| CN107958235A (en) * | 2017-12-28 | 2018-04-24 | 泰康保险集团股份有限公司 | A kind of facial image detection method, device, medium and electronic equipment |
| CN108133391A (en) * | 2017-12-22 | 2018-06-08 | 联想(北京)有限公司 | Method for Sales Forecast method and server |
| CN108388291A (en) * | 2018-01-17 | 2018-08-10 | 中国农业大学 | A kind of greenhouse cluster environment regulation and control method and system |
| CN108446771A (en) * | 2018-04-02 | 2018-08-24 | 四川长虹电器股份有限公司 | A method of preventing Sale Forecasting Model over-fitting |
-
2018
- 2018-11-09 CN CN201811334201.XA patent/CN109447716A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105678588A (en) * | 2016-01-13 | 2016-06-15 | 焦点科技股份有限公司 | Product sales volume fluctuation assessment method based on e-commerce platform |
| CN107239532A (en) * | 2017-05-31 | 2017-10-10 | 北京京东尚科信息技术有限公司 | Data digging method and device |
| CN108133391A (en) * | 2017-12-22 | 2018-06-08 | 联想(北京)有限公司 | Method for Sales Forecast method and server |
| CN107958235A (en) * | 2017-12-28 | 2018-04-24 | 泰康保险集团股份有限公司 | A kind of facial image detection method, device, medium and electronic equipment |
| CN108388291A (en) * | 2018-01-17 | 2018-08-10 | 中国农业大学 | A kind of greenhouse cluster environment regulation and control method and system |
| CN108446771A (en) * | 2018-04-02 | 2018-08-24 | 四川长虹电器股份有限公司 | A method of preventing Sale Forecasting Model over-fitting |
Non-Patent Citations (2)
| Title |
|---|
| 宗晓杰 编著: "《基于数据挖掘的农机销售预测模型研究》", 30 November 2009, 北京：中国科学技术出版社 * |
| 曾繁慧 等: "因素分析法的样本培育"", 《辽宁工程技术大学学报(自然科学版)》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110458664A (en) * | 2019-08-06 | 2019-11-15 | 上海新共赢信息科技有限公司 | A kind of user's trip information prediction technique, device, equipment and storage medium |
| CN111291657A (en) * | 2020-01-21 | 2020-06-16 | 同济大学 | Crowd counting model training method based on difficult case mining and application |
| CN111652655A (en) * | 2020-06-10 | 2020-09-11 | 创新奇智(上海)科技有限公司 | Commodity sales prediction method and device, electronic equipment and readable storage medium |
| CN113837783A (en) * | 2020-06-24 | 2021-12-24 | 上海顺如丰来技术有限公司 | Time series model parameter optimization method and device and computer equipment |
| CN113283936A (en) * | 2021-05-28 | 2021-08-20 | 深圳千岸科技股份有限公司 | Sales forecasting method, sales forecasting device and electronic equipment |
| CN113763186A (en) * | 2021-10-22 | 2021-12-07 | 平安科技（深圳）有限公司 | User transfer insurance prediction method, device and equipment based on recurrent neural network |
| CN113763186B (en) * | 2021-10-22 | 2024-03-15 | 平安科技（深圳）有限公司 | User transfer prediction method, device and equipment based on cyclic neural network |
| CN114581157A (en) * | 2022-04-28 | 2022-06-03 | 湖南康道医药有限公司 | Sales prediction method and apparatus based on big data, electronic device, and medium |
| CN114581157B (en) * | 2022-04-28 | 2022-11-04 | 湖南康道医药有限公司 | Sales volume prediction method and device based on big data, electronic equipment and medium |
| CN115829630A (en) * | 2022-12-16 | 2023-03-21 | 广州飞狮数字科技有限公司 | Method and device for determining sales ratio of shop |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109447716A (en) | Method for Sales Forecast method and server based on Recognition with Recurrent Neural Network | |
| Cattaneo et al. | A practical introduction to regression discontinuity designs: Foundations | |
| CN106781489B (en) | A kind of road network trend prediction method based on recurrent neural network | |
| Laslett et al. | A flexible maximum likelihood approach for fitting growth curves to tag recapture data | |
| Bernard et al. | Using artificial neural networks to identify learning styles | |
| CN110378052A (en) | It is looked to the future the equipment method for predicting residual useful life of operating condition based on Recognition with Recurrent Neural Network | |
| CN108254741A (en) | Targetpath Forecasting Methodology based on Recognition with Recurrent Neural Network | |
| Carlson et al. | Biologically plausible models of homeostasis and STDP: stability and learning in spiking neural networks | |
| CN109800801A (en) | K-Means clustering lane method of flow based on Gauss regression algorithm | |
| CN109360604A (en) | A kind of oophoroma molecule parting forecasting system | |
| CN106504535B (en) | A kind of trip distribution modeling method of combination Gravity Models and Fratar models | |
| CN106570516A (en) | Obstacle recognition method using convolution neural network | |
| CN109544196A (en) | The price multiple of reservation order determines method and apparatus | |
| CN102541920A (en) | Method and device for improving accuracy degree by collaborative filtering jointly based on user and item | |
| CN109615760A (en) | Intelligently add the method and system of clear paper money to ATM in bank based on machine learning | |
| CN103942461A (en) | Water quality parameter prediction method based on online sequential extreme learning machine | |
| CN107085941A (en) | A kind of traffic flow forecasting method, apparatus and system | |
| CN110289081A (en) | The epilepsy detection method of depth network stack model adaptation weighted feature fusion | |
| CN107481218A (en) | Image aesthetic feeling appraisal procedure and device | |
| CN109451304A (en) | A kind of camera module batch focusing test method and system | |
| CN107945534A (en) | A kind of special bus method for predicting based on GMDH neutral nets | |
| CN106251260A (en) | A kind of candidates' aspiration makes a report on analog systems and method | |
| CN107038492A (en) | Daily Order volume Forecasting Methodology and device based on Arma models | |
| CN109190810B (en) | TDNN-based prediction method for NDVI (normalized difference vegetation index) of northern grassland area of China | |
| CN108446771A (en) | A method of preventing Sale Forecasting Model over-fitting |
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: 20190308 |