CN116911908A - Sales data prediction method and system based on artificial intelligence - Google Patents
Sales data prediction method and system based on artificial intelligence Download PDFInfo
- Publication number
- CN116911908A CN116911908A CN202310911626.7A CN202310911626A CN116911908A CN 116911908 A CN116911908 A CN 116911908A CN 202310911626 A CN202310911626 A CN 202310911626A CN 116911908 A CN116911908 A CN 116911908A
- Authority
- CN
- China
- Prior art keywords
- sales
- product
- data
- seasonal
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 25
- 230000001932 seasonal effect Effects 0.000 claims abstract description 93
- 238000013528 artificial neural network Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims description 147
- 230000004927 fusion Effects 0.000 claims description 43
- 238000002372 labelling Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 6
- 239000003086 colorant Substances 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 abstract description 4
- 238000013527 convolutional neural network Methods 0.000 description 5
- 238000000605 extraction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- 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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Business, Economics & Management (AREA)
- Databases & Information Systems (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- Accounting & Taxation (AREA)
- Multimedia (AREA)
- Finance (AREA)
- Medical Informatics (AREA)
- Strategic Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a sales data prediction method, a sales data prediction system and electronic equipment based on artificial intelligence, wherein five sales background data are obtained; obtaining a sales graph according to the sales geographic data; and obtaining a plurality of sales volumes through a sales relationship network according to the product coefficients, the five sales figures and the seasonal coefficients. And adopting five consecutive sales background data to consider the time relation, and predicting sales data according to the time change condition. According to the method, the geographical features of sales data distributed on geographical positions can be more accurately considered according to the form of converting sales background data into sales diagrams, and according to the seasonal factors and the product factors, the seasonal features can be added to time features and the product features can be added to the geographical features when the neural network predicts, so that sales can be more accurately predicted.
Description
Technical Field
The invention relates to the technical field of computers, in particular to an artificial intelligence-based sales data prediction method and system.
Background
Since sales data is large and the sales data has regularity in time, region, and the like, the sales data can be predicted by an artificial intelligence method. However, since the extraction of the time-neural network is generally adopted for the extraction of the characteristics of the sales data at present, the time-neural network is not specific to the geographic position, and the time relationship, the seasonal relationship and the product relationship are not considered. So that the prediction is not accurate enough.
Disclosure of Invention
The invention aims to provide an artificial intelligence-based sales data prediction method and system, which are used for solving the problems in the prior art.
In a first aspect, an embodiment of the present invention provides an artificial intelligence-based sales data prediction method, including:
five sales background data are obtained; the sales context data includes season, sales geographic data, and product context data; the sales geographic data comprises sales date, sales area, sales location, dealer and channel classification; the product background data comprises a product serial number, a product category and a product price;
obtaining a sales graph according to the sales geographic data; five sales background data are correspondingly obtained to obtain five sales figures, and the sales background data are in one-to-one correspondence with the sales figures; the sales map is a two-dimensional image marked with sales background data;
obtaining a product coefficient according to the product serial number and the product sales relationship;
according to the seasons, obtaining seasonal coefficients through a seasonal sales relationship;
obtaining a plurality of sales volumes through a sales relationship network according to the product coefficients, the five sales figures and the seasonal coefficients; one product category corresponds to one sales volume;
the sales relationship network comprises a two-dimensional convolution network, a three-dimensional convolution network and a fusion network; the converged network comprises a first converged network, a second converged network and a third converged network.
Optionally, the training method of the sales relationship network includes:
obtaining a training set; the training set comprises a plurality of training images and a plurality of annotation data; the training image is a binary map marked with historical sales background data; the labeling data represents labeling product categories and labeling sales; the marked product category represents categories of a plurality of products sold at a time point; the marked sales amount represents the number of products sold corresponding to the marked product category;
obtaining a seasonal sales relationship based on the training set; the seasonal sales relationship comprises a spring sales relationship, a summer sales relationship, an autumn sales relationship and a winter sales relationship;
obtaining a product sales relationship based on the training set; the product sales relationship comprises a plurality of product types and sales relationships;
dividing a plurality of training images according to the time sequence and the step number of 5 to obtain 5 training images;
inputting the 5 training images into a two-dimensional convolution network, judging image data, and obtaining a first output set; the first output set includes a plurality of first output values;
inputting the 5 training images into a three-dimensional convolution network, and judging a time relationship to obtain a second output value;
based on the first output set, the second output value, the seasonal sales relationship and the product sales relationship, obtaining a predicted historical sales volume through a fused neural network;
and after the predicted historical sales and the marked sales calculate the loss value, the parameters of the sales relation network are back-propagated and trained.
Optionally, obtaining the seasonal sales relationship based on the training set includes:
classifying the training sets according to seasons to obtain four training image sets; the four training image sets comprise a spring pin training image set, a summer training image set, an autumn training image set and a winter training image set; the training image set represents training images segmented according to seasons;
according to the four training image sets, respectively acquiring data with the occurrence times of historical sales background data in the sets being greater than that of other data to obtain four repeated training image sets;
according to the four repeated training image sets, performing similarity judgment on images in the repeated training image sets to obtain four similar time sets; the value in the similar time set is the sales time corresponding to the high similar image in the repeated training image set; the high similarity image represents an image having a greater number of similarity than other images;
adding sales corresponding to the similar time sets to obtain total seasonal sales; four total seasonal sales correspond to four sets of similar times;
dividing the four total seasonal sales by the sum of the four total seasonal sales to obtain four seasonal coefficients; one seasonal factor corresponds to one total seasonal sales amount corresponding to one season;
and constructing an association relation between the seasonal coefficient and the season to obtain a seasonal sales relation.
Optionally, the inputting the 5 training images into a two-dimensional convolution network, judging image data, and obtaining a first output set includes:
judging the similarity between every two training images according to the 5 training images to obtain a plurality of similar two-dimensional image sets; training images in the similar two-dimensional image set are similar;
randomly acquiring similar two-dimensional images from the similar two-dimensional image set; the plurality of similar two-dimensional image sets correspond to the plurality of similar two-dimensional images;
and respectively inputting the plurality of similar two-dimensional images into a two-dimensional convolution network, and detecting different pixels and distribution conditions of the pixels in the images to obtain a plurality of first output values.
Optionally, the inputting the 5 training images into a three-dimensional convolution network, judging a time relationship, and obtaining a second output value includes:
overlapping the 5 training images according to the time sequence to obtain a training three-dimensional image;
inputting the training three-dimensional image into a three-dimensional convolution network to obtain second training data;
the number of the first dimension of the three-dimensional convolution kernels of the three-dimensional convolution network is smaller than the length of the training image; the number of the second dimension of the convolution kernels of the three-dimensional convolution network is smaller than the width of the training image; the number of the third dimension of the convolution kernel of the three-dimensional convolution network is 5;
the number of steps of the three-dimensional convolution kernel of the three-dimensional convolution network is 1.
Optionally, the obtaining predicted sales data by fusing the neural network based on the first output set, the second output value, the seasonal sales relationship, and the product sales relationship includes:
obtaining a first product coefficient; the first product coefficient is a product coefficient corresponding to the first output value in the product sales relationship;
inputting a plurality of first output values in the first output set and the first product coefficients into a first fusion network to obtain a first fusion value;
the corresponding seasonal coefficient in the seasonal sales relation is used as a first seasonal coefficient according to the second output value;
inputting the second output value and the first seasonal coefficient into a second fusion network to obtain a second fusion value;
and inputting the first fusion value and the second fusion value into a third fusion network to obtain predicted historical sales data.
Optionally, the obtaining a sales graph according to the sales geographic data includes:
obtaining a binary map; the binary map is a binary image marked with city division;
marking a sales area for selling geographic data in the binary map with black;
marking the selling places of the selling geographic data in the binary map with red points;
marking green points of dealers selling geographic data in the binary map;
connecting points of the dealer and the sales place in the binary map through curves with different colors to serve as sales channels;
and mapping the marked binary map with the selling date of the selling geographic data one by one to obtain a selling map.
Optionally, the obtaining the product sales relationship based on the training set includes:
adding the marked sales volumes corresponding to the marked product categories to obtain total sales volume;
dividing the marked sales volume corresponding to the marked product category by the total sales volume to obtain a product coefficient; the marked product category corresponds to a plurality of product coefficients;
and constructing an association relation between the product coefficients and the marked product categories to obtain a product sales relation.
Optionally, the obtaining a plurality of sales volumes through a sales relationship network according to the product coefficients, the five sales figures and the seasonal coefficients includes:
the input of the two-dimensional convolution network is a sales graph; the input of the three-dimensional convolution network is a sales graph; the input of the first fusion network is the output of a seasonal factor and a two-dimensional convolution network; the input of the second fusion network is the output of the product coefficient and the three-dimensional convolution network; the input of the third fusion network is the output of the first fusion network and the output of the second fusion network;
and predicting the seasonal factors, the product factors and the sales figures according to the same method as the training sales relation network to obtain predicted sales data.
In a second aspect, an embodiment of the present invention provides an artificial intelligence based sales data prediction system, including:
the acquisition module is used for: five sales background data are obtained; the sales context data includes season, sales geographic data, and product context data; the sales geographic data comprises sales date, sales area, sales location, dealer and channel classification; the product background data comprises a product serial number, a product category and a product price;
and a composition module: obtaining a sales graph according to the sales geographic data; five sales background data are correspondingly obtained to obtain five sales figures, and the sales background data are in one-to-one correspondence with the sales figures; the sales map is a two-dimensional image marked with sales background data;
and a product coefficient module: obtaining a product coefficient according to the product serial number and the product sales relationship;
and a seasonal coefficient module: according to the seasons, obtaining seasonal coefficients through a seasonal sales relationship;
and a prediction module: obtaining a plurality of sales volumes through a sales relationship network according to the product coefficients, the five sales figures and the seasonal coefficients; one product category corresponds to one sales volume;
the sales relationship network comprises a two-dimensional convolution network, a three-dimensional convolution network and a fusion network; the converged network comprises a first converged network, a second converged network and a third converged network.
Compared with the prior art, the embodiment of the invention achieves the following beneficial effects:
the embodiment of the invention also provides a sales data prediction method and a sales data prediction system based on artificial intelligence, wherein the method comprises the following steps: five sales background data are obtained; the sales context data includes season, sales geographic data, and product context data; the sales geographic data comprises sales date, sales area, sales location, dealer and channel classification; the product background data comprises a product serial number, a product category and a product price; obtaining a sales graph according to the sales geographic data; five sales background data are correspondingly obtained to obtain five sales figures, and the sales background data are in one-to-one correspondence with the sales figures; the sales map is a two-dimensional image marked with sales background data; obtaining a product coefficient according to the product serial number and the product sales relationship; according to the seasons, obtaining seasonal coefficients through a seasonal sales relationship; obtaining a plurality of sales volumes through a sales relationship network according to the product coefficients, the five sales figures and the seasonal coefficients; one product category corresponds to one sales volume; the sales relationship network comprises a two-dimensional convolution network, a three-dimensional convolution network and a fusion network; the converged network comprises a first converged network, a second converged network and a third converged network.
And adopting five consecutive sales background data to consider the time relation, and predicting sales data according to the time change condition. According to the method, the geographical features of sales data distributed on geographical positions can be more accurately considered according to the form of converting sales background data into sales diagrams, and according to the seasonal factors and the product factors, the seasonal features can be added to time features and the product features can be added to the geographical features when the neural network predicts, so that sales can be more accurately predicted.
Drawings
FIG. 1 is a flowchart of a sales data prediction method based on artificial intelligence according to an embodiment of the present invention.
Fig. 2 is a schematic block diagram of an electronic device according to an embodiment of the present invention.
The marks in the figure: a bus 500; a receiver 501; a processor 502; a transmitter 503; a memory 504; bus interface 505.
Detailed Description
The present invention will be described in detail with reference to the accompanying drawings.
Examples
As shown in fig. 1, an embodiment of the present invention provides a sales data prediction method based on artificial intelligence, which includes:
s101: five sales background data are obtained; the sales context data includes season, sales geographic data, and product context data; the sales geographic data comprises sales date, sales area, sales location, dealer and channel classification; the product background data comprises a product serial number, a product category and a product price;
s102: obtaining a sales graph according to the five sales background data; five sales background data are correspondingly obtained to obtain five sales figures, and the sales background data are in one-to-one correspondence with the sales figures; the sales map is a two-dimensional image marked with sales background data;
the lowest layer of the sales map is a two-dimensional binary map with the sales background data marked and the time farthest from the current time. The uppermost layer of the sales map is a two-dimensional binary map with the time being the farthest from the current time and marked with sales background data.
S103: obtaining a product coefficient according to the product serial number and the product sales relationship;
s104: according to the seasons, obtaining seasonal coefficients through sales and seasonal relations;
s105: obtaining a plurality of sales volumes through a sales relationship network according to the product coefficients, the five sales figures and the seasonal coefficients; one product category corresponds to one sales volume.
The corresponding product price can be obtained through the association relation of the product categories. The sales are obtained by multiplying the product price and sales. Wherein, the product category and the product price have established an association relationship.
The sales relationship network comprises a two-dimensional convolution network, a three-dimensional convolution network and a fusion network; the converged network comprises a first converged network, a second converged network and a third converged network.
Optionally, the training method of the sales relationship network includes:
obtaining a training set; the training set comprises a plurality of training images and a plurality of annotation data; the training image is a binary map marked with historical sales background data; the labeling data represents labeling product categories and labeling sales; the marked product category represents categories of a plurality of products sold at a time point; the marked sales amount indicates the number of products sold corresponding to the marked product category.
Wherein one training image corresponds to one annotation data.
The training image is a sales image formed by sales data corresponding to a plurality of time points of a plurality of years of history.
And obtaining the seasonal sales relationship based on the training set. Spring sales, summer sales, autumn sales and winter sales.
The spring sales relationship represents a state of sales in spring, the summer sales relationship represents a state of sales in summer, the autumn sales relationship represents a state of sales in autumn, and the winter sales relationship represents a state of sales in a dynamic diagram.
Obtaining a product sales relationship based on the training set; the product sales relationship includes a plurality of product types versus sales.
And dividing the training images by 5 steps according to the time sequence to obtain 5 training images.
Wherein the time sequence of the 5 training images is adjacent to each other. The first training image is an image of 1 month and 1 day of sales, the second training image is an image of 1 month and 2 days of sales, the third training image is an image of 1 month and 3 days of sales, the fourth training image is an image of 1 month and 4 days of sales, and the fifth training image is an image of 1 month and 5 days of sales.
Wherein, if 365 days of a year there is one image on sales every day, then the image on sales of the past 1 year is taken as a training image. The segmentation was performed with a step number of 5, and 73 training images for training were obtained.
Inputting the 5 training images into a two-dimensional convolution network, judging image data, and obtaining a first output set; the first output set includes a plurality of first output values.
Wherein the first output value is an output value of a two-dimensional convolution network.
And inputting the training data into a three-dimensional convolution network, and judging the time relationship to obtain a second output value.
The second output value is an output value of a three-dimensional convolution network.
Based on the first output set, the second output value, the seasonal sales relationship and the product sales relationship, obtaining a predicted historical sales volume through a fused neural network;
and after the predicted historical sales and the marked sales calculate the loss value, the parameters of the sales relation network are back-propagated and trained.
And calculating the loss values of the predicted historical sales and the marked sales through a cross entropy loss function.
By the method, the trend of a large amount of data is judged according to the training data corresponding to the marked data, the seasonal sales relationship and the product sales relationship are found, and the predicted historical sales data are also found.
By the method, sales can be affected in different seasons, different times, different sales channels, different sales products and the like, so that various aspects need to be accurately considered.
Optionally, obtaining the seasonal sales relationship based on the training set includes:
classifying the training sets according to seasons to obtain four training image sets; the four training image sets comprise a spring pin training image set, a summer training image set, an autumn training image set and a winter training image set; the training image set represents a training image segmented according to seasons.
Wherein the training image set also represents sales background data of a history of segmentation according to seasons. The historical sales background data marks a training image in a retraining image set.
And respectively acquiring data with the occurrence times of the historical sales background data in the sets being greater than that of other data according to the four training image sets to obtain four repeated training image sets.
The four repeated training image sets are a spring repeated training image set, a summer repeated training image set, an autumn repeated training image set and a winter repeated training image set.
And if the number of times of occurrence of the marked sales areas in the training image set is larger than the number of times of occurrence of other sales areas, inputting the training images corresponding to the sales areas into the repeated training image set. And inputting training images corresponding to the most frequently-occurring sales locations, distributors and channel classifications into the repeated training image set after a plurality of times.
And carrying out similarity judgment on images in the repeated training image sets according to the four repeated training image sets to obtain four similar time sets. The value in the similar time set is the sales time corresponding to the high similar image in the repeated training image set; the high similarity image represents an image having a greater number of similarities than the other images.
And the value in the similar time set is a set formed by selling the most similar images in the repeated training image set.
In the present embodiment, pixel values of two images are compared one by one through Mean Square Error (MSE), and the mean square error is calculated and used for judging the similarity of the images in the set, so that the similar images are stored and purchased in the set to obtain a similar time set. In this embodiment, the threshold of the mean square error is set to 2, which means that when two images in the repeated training image set calculate the mean square error, the two images are similar when the mean square error is smaller than 2, and the two images are dissimilar when the mean square error is larger than 2.
Wherein a set of similar times corresponds to a set of repeated training images.
The four similar time sets comprise a spring similar time set, a summer similar time set, an autumn similar time set and a winter similar time set, and correspond to a spring repeated training image set, a summer repeated training image set, an autumn repeated training image set and a winter repeated data set respectively.
Adding sales corresponding to the similar time sets to obtain total seasonal sales; four total seasonal sales correspond to four sets of similar times;
dividing the four total seasonal sales by the sum of the four total seasonal sales to obtain four seasonal coefficients; one seasonal factor corresponds to one total seasonal sales and one season.
And constructing an association relation between the seasonal coefficient and the season to obtain a seasonal sales relation.
By the method, the reason for judging the seasonal sales relationship is as follows: each season must affect the sales of one product, finding the relationship between the season and the sales corresponding to each product.
Optionally, the inputting the 5 training images into a two-dimensional convolution network, judging image data, and obtaining a first output value includes:
judging the similarity between every two training images according to the 5 training images to obtain a plurality of similar two-dimensional image sets; training images in the set of similar two-dimensional images are similar.
In the present embodiment, pixel values of two images are compared one by one through Mean Square Error (MSE), and the mean square error is calculated and used for judging the similarity of the images in the set, so that the similar images are stored and purchased in the set to obtain a similar time set. In this embodiment, the threshold of the mean square error is set to 2, which means that when two images in the repeated training image set calculate the mean square error, the two images are similar when the mean square error is smaller than 2, and the two images are dissimilar when the mean square error is larger than 2.
Randomly acquiring similar two-dimensional images from the similar two-dimensional image set; the plurality of similar two-dimensional image sets correspond to the plurality of similar two-dimensional images;
and respectively inputting the plurality of similar two-dimensional images into a two-dimensional convolution network, and detecting different pixels and distribution conditions of the pixels in the images to obtain a plurality of first output values.
Wherein the two-dimensional convolutional network is a convolutional neural network (Convolutional Neural Networks, CNN). In this embodiment, the output layer of the two-dimensional convolutional network is 1024.
Wherein the two-dimensional convolution network comprises a plurality of two-dimensional convolution kernels; the number of the two-dimensional convolution kernels is a multiple of the number of the training images. The step length of the two-dimensional convolution network is 1.
Optionally, the inputting the 5 training images into a three-dimensional convolution network, judging a time relationship, and obtaining a second output value includes:
overlapping the 5 training images according to the time sequence to obtain a training three-dimensional image;
and inputting the training three-dimensional image into a three-dimensional convolution network to obtain second training data.
Wherein the three-dimensional convolutional network is a three-dimensional convolutional neural network (3D Convolutional Neural Networks,3DCNN).
The number of the first dimension of the three-dimensional convolution kernels of the three-dimensional convolution network is smaller than the length of the training image; the number of the second dimension of the convolution kernels of the three-dimensional convolution network is smaller than the width of the training image; the number of the third dimension of the convolution kernel of the three-dimensional convolution network is 5;
the number of steps of the three-dimensional convolution kernel of the three-dimensional convolution network is 1.
By the above method, information about time is extracted with a three-dimensional convolution network.
Optionally, the obtaining predicted sales data by fusing the neural network based on the first output set, the second output value, the seasonal sales relationship, and the product sales relationship includes:
obtaining a first product coefficient; the first product coefficient is a product coefficient corresponding to the first output value in the product sales relationship;
inputting a plurality of first output values in the first output set and the first product coefficients into a first fusion network to obtain a first fusion value;
the corresponding seasonal coefficient in the seasonal sales relation is used as a first seasonal coefficient according to the second output value;
inputting the second output value and the first seasonal coefficient into a second fusion network to obtain a second fusion value;
and inputting the first fusion value and the second fusion value into a third fusion network to obtain predicted historical sales data.
The first fusion network, the first fusion network and the first fusion network are deep neural networks (Deep Neural Networks) with different numbers of neurons and layers.
The number of output layers of the third fusion network is the number of product categories.
By means of the method, the seasonal coefficients of different seasons are fused in time, the sales geographic images are fused with the product characteristics, and sales corresponding to the product types can be accurately predicted according to the seasons, the time, the products and the geographical characteristics of sales.
Optionally, the obtaining a sales graph according to the sales geographic data includes:
obtaining a binary map; the binary map is a binary image marked with city division;
and marking the selling area selling the geographic data in the binary map with black.
Wherein the selling area of the geographic data in the binary map is black blocks.
Marking the selling places of the selling geographic data in the binary map with red points;
marking green points of dealers selling geographic data in the binary map;
and connecting the point of the dealer in the binary map with the point of the sales place by curves with different colors, and taking the point as a sales channel.
Different points, lines and areas are marked by different colors and different RGB values and are used for reflecting different sales geographic data.
And mapping the marked binary map with the selling date of the selling geographic data one by one to obtain a selling map.
By the method, sales amount is different according to the number, the type and the cost of the products, different distributors, the number of the sold products, different channels and the cost of the road loss. Various aspects are considered and embodied in the form of geographic images are provided.
Optionally, the obtaining the product coefficient according to the product background data and the product sales relationship includes:
adding the marked sales volumes corresponding to the marked product categories to obtain total sales volume;
dividing the marked sales volume corresponding to the marked product category by the total sales volume to obtain a product coefficient; the noted product category corresponds to a plurality of product coefficients.
Wherein, the marked product category and the marked sales volume have constructed a one-to-one association relation.
And constructing an association relation between the product coefficients and the marked product categories to obtain a product sales relation.
Optionally, the obtaining sales data according to the product coefficient, the sales map and the seasonal coefficient through a sales relationship network includes:
the input of the two-dimensional convolution network is a sales graph; the input of the three-dimensional convolution network is a sales graph; the input of the first fusion network is the output of a seasonal factor and a two-dimensional convolution network; the input of the second fusion network is the output of the product coefficient and the three-dimensional convolution network; the input of the third fusion network is the output of the first fusion network and the output of the second fusion network;
and predicting the seasonal factors, the product factors and the sales figures according to the same method as the training sales relation network to obtain predicted sales data.
Examples
Based on the sales data prediction method based on the artificial intelligence, the embodiment of the invention also provides a sales data prediction system based on the artificial intelligence, which comprises an acquisition module, a composition module, a product coefficient module, a seasonal coefficient module and a prediction module.
The acquisition module is used for acquiring five sales background data; the sales context data includes season, sales geographic data, and product context data; the sales geographic data comprises sales date, sales area, sales location, dealer and channel classification; the product background data comprises a product serial number, a product category and a product price;
the composition module is used for obtaining a sales graph according to the sales geographic data; five sales background data are correspondingly obtained to obtain five sales figures, and the sales background data are in one-to-one correspondence with the sales figures; the sales map is a two-dimensional image marked with sales background data;
the product coefficient module is used for obtaining a product coefficient according to the product serial number and the product sales relationship;
the seasonal coefficient module is used for obtaining a seasonal coefficient according to the seasons and the seasonal sales relation;
the prediction module is used for obtaining a plurality of sales volumes through a sales relationship network according to the product coefficients, the five sales figures and the seasonal coefficients; one product category corresponds to one sales volume;
the sales relationship network comprises a two-dimensional convolution network, a three-dimensional convolution network and a fusion network; the converged network comprises a first converged network, a second converged network and a third converged network.
An embodiment of the present invention further provides an electronic device, as shown in fig. 2, including a memory 504, a processor 502, and a computer program stored in the memory 504 and executable on the processor 502, where the processor 502 implements the steps of any one of the above-mentioned sales data prediction methods based on artificial intelligence when executing the program.
Where in FIG. 2 a bus architecture (represented by bus 500), bus 500 may include any number of interconnected buses and bridges, with bus 500 linking together various circuits, including one or more processors, represented by processor 502, and memory, represented by memory 504. Bus 500 may also link together various other circuits such as peripheral devices, voltage regulators, power management circuits, etc., as are well known in the art and, therefore, will not be described further herein. Bus interface 505 provides an interface between bus 500 and receiver 501 and transmitter 503. The receiver 501 and the transmitter 503 may be the same element, i.e. a transceiver, providing a means for communicating with various other apparatus over a transmission medium. The processor 502 is responsible for managing the bus 500 and general processing, while the memory 504 may be used to store data used by the processor 502 in performing operations.
Embodiments of the present invention also provide a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of any of the methods of an artificial intelligence based sales data prediction method described above, as well as the data referred to above.
The algorithms and displays presented herein are not inherently related to any particular computer, virtual system, or other apparatus. Various general-purpose systems may also be used with the teachings herein. The required structure for a construction of such a system is apparent from the description above. In addition, the present invention is not directed to any particular programming language. It will be appreciated that the teachings of the present invention described herein may be implemented in a variety of programming languages, and the above description of specific languages is provided for disclosure of enablement and best mode of the present invention.
In the description provided herein, numerous specific details are set forth. However, it is understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.
Similarly, it should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be construed as reflecting the intention that: i.e., the claimed invention requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.
Those skilled in the art will appreciate that the modules in the apparatus of the embodiments may be adaptively changed and disposed in one or more apparatuses different from the embodiments. The modules or units or components of the embodiments may be combined into one module or unit or component and, furthermore, they may be divided into a plurality of sub-modules or sub-units or sub-components. Any combination of all features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or units of any method or apparatus so disclosed, may be used in combination, except insofar as at least some of such features and/or processes or units are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings), may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.
Furthermore, those skilled in the art will appreciate that while some embodiments herein include some features but not others included in other embodiments, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments can be used in any combination.
Various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that some or all of the functions of some or all of the components in an apparatus according to embodiments of the present invention may be implemented in practice using a microprocessor or Digital Signal Processor (DSP). The present invention can also be implemented as an apparatus or device program (e.g., a computer program and a computer program product) for performing a portion or all of the methods described herein. Such a program embodying the present invention may be stored on a computer readable medium, or may have the form of one or more signals. Such signals may be downloaded from an internet website, provided on a carrier signal, or provided in any other form.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the words first, second, third, etc. do not denote any order. These words may be interpreted as names.
Claims (10)
1. An artificial intelligence based sales data prediction method, comprising:
five sales background data are obtained; the sales context data includes season, sales geographic data, and product context data; the sales geographic data comprises sales date, sales area, sales location, dealer and channel classification; the product background data comprises a product serial number, a product category and a product price;
obtaining a sales graph according to the sales geographic data; five sales background data are correspondingly obtained to obtain five sales figures, and the sales background data are in one-to-one correspondence with the sales figures; five sales background data are correspondingly obtained to obtain five sales figures, and the sales background data are in one-to-one correspondence with the sales figures; the sales map is a two-dimensional image marked with sales background data;
obtaining a product coefficient according to the product serial number and the product sales relationship;
according to the seasons, obtaining seasonal coefficients through a seasonal sales relationship;
obtaining a plurality of sales volumes through a sales relationship network according to the product coefficients, the five sales figures and the seasonal coefficients; one product category corresponds to one sales volume;
the sales relationship network comprises a two-dimensional convolution network, a three-dimensional convolution network and a fusion network; the converged network comprises a first converged network, a second converged network and a third converged network.
2. The method for predicting sales data based on artificial intelligence of claim 1, wherein the training method for the sales relationship network comprises:
obtaining a training set; the training set comprises a plurality of training images and a plurality of annotation data; the training image is a binary map marked with historical sales background data; the labeling data represents labeling product categories and labeling sales; the marked product category represents categories of a plurality of products sold at a time point; the marked sales amount represents the number of products sold corresponding to the marked product category;
obtaining a seasonal sales relationship based on the training set; the seasonal sales relationship comprises a spring sales relationship, a summer sales relationship, an autumn sales relationship and a winter sales relationship;
obtaining a product sales relationship based on the training set; the product sales relationship comprises a plurality of product types and sales relationships;
dividing a plurality of training images according to the time sequence and the step number of 5 to obtain 5 training images;
inputting the 5 training images into a two-dimensional convolution network, judging image data, and obtaining a first output set; the first output set includes a plurality of first output values;
inputting the 5 training images into a three-dimensional convolution network, and judging a time relationship to obtain a second output value;
based on the first output set, the second output value, the seasonal sales relationship and the product sales relationship, obtaining a predicted historical sales volume through a fused neural network;
and after the predicted historical sales and the marked sales calculate the loss value, the parameters of the sales relation network are back-propagated and trained.
3. The artificial intelligence based sales data prediction method according to claim 2, wherein the sales data prediction method is based on the following
Training set, obtain seasonal sales relationship, including:
classifying the training sets according to seasons to obtain four training image sets; the four training image sets comprise a spring pin training image set, a summer training image set, an autumn training image set and a winter training image set; the training image set represents training images segmented according to seasons;
according to the four training image sets, respectively acquiring data with the occurrence times of historical sales background data in the sets being greater than that of other data to obtain four repeated training image sets;
according to the four repeated training image sets, performing similarity judgment on images in the repeated training image sets to obtain four similar time sets; the value in the similar time set is the sales time corresponding to the high similar image in the repeated training image set; the high similarity image represents an image having a greater number of similarity than other images;
adding sales corresponding to the similar time sets to obtain total seasonal sales; four total seasonal sales correspond to four sets of similar times;
dividing the four total seasonal sales by the sum of the four total seasonal sales to obtain four seasonal coefficients; one seasonal factor corresponds to one total seasonal sales amount corresponding to one season;
and constructing an association relation between the seasonal coefficient and the season to obtain a seasonal sales relation.
4. The sales data prediction method based on artificial intelligence according to claim 2, wherein the inputting the 5 training images into a two-dimensional convolution network, judging the image data, and obtaining a first output set includes:
judging the similarity between every two training images according to the 5 training images to obtain a plurality of similar two-dimensional image sets; training images in the similar two-dimensional image set are similar;
randomly acquiring similar two-dimensional images from the similar two-dimensional image set; the plurality of similar two-dimensional image sets correspond to the plurality of similar two-dimensional images;
and respectively inputting the plurality of similar two-dimensional images into a two-dimensional convolution network, and detecting different pixels and distribution conditions of the pixels in the images to obtain a plurality of first output values.
5. The sales data prediction method based on artificial intelligence according to claim 2, wherein the inputting the 5 training images into the three-dimensional convolution network, determining the time relationship, and obtaining the second output value includes:
overlapping the 5 training images according to the time sequence to obtain a training three-dimensional image;
inputting the training three-dimensional image into a three-dimensional convolution network to obtain second training data;
the number of the first dimension of the three-dimensional convolution kernels of the three-dimensional convolution network is smaller than the length of the training image; the number of the second dimension of the convolution kernels of the three-dimensional convolution network is smaller than the width of the training image; the number of the third dimension of the convolution kernel of the three-dimensional convolution network is 5;
the number of steps of the three-dimensional convolution kernel of the three-dimensional convolution network is 1.
6. The artificial intelligence based sales data prediction method according to claim 2, wherein the artificial intelligence based sales data prediction method comprises the steps of
The first output set, the second output value, the seasonal sales relationship and the product sales relationship are fused to obtain predicted sales data through a fusion neural network, and the method comprises the following steps:
obtaining a first product coefficient; the first product coefficient is a product coefficient corresponding to the first output value in the product sales relationship;
inputting a plurality of first output values in the first output set and the first product coefficients into a first fusion network to obtain a first fusion value;
the corresponding seasonal coefficient in the seasonal sales relation is used as a first seasonal coefficient according to the second output value;
inputting the second output value and the first seasonal coefficient into a second fusion network to obtain a second fusion value;
and inputting the first fusion value and the second fusion value into a third fusion network to obtain predicted historical sales data.
7. The method for predicting sales data based on artificial intelligence according to claim 1, wherein the obtaining a sales map based on sales geographic data comprises:
obtaining a binary map; the binary map is a binary image marked with city division;
marking a sales area for selling geographic data in the binary map with black;
marking the selling places of the selling geographic data in the binary map with red points;
marking green points of dealers selling geographic data in the binary map;
connecting points of the dealer and the sales place in the binary map through curves with different colors to serve as sales channels;
and mapping the marked binary map with the selling date of the selling geographic data one by one to obtain a selling map.
8. The sales data prediction method based on artificial intelligence according to claim 2, wherein the obtaining the product sales relationship based on the training set comprises:
adding the marked sales volumes corresponding to the marked product categories to obtain total sales volume;
dividing the marked sales volume corresponding to the marked product category by the total sales volume to obtain a product coefficient; the marked product category corresponds to a plurality of product coefficients;
and constructing an association relation between the product coefficients and the marked product categories to obtain a product sales relation.
9. An artificial intelligence based sales data prediction system, comprising:
the acquisition module is used for: five sales background data are obtained; the sales context data includes season, sales geographic data, and product context data; the sales geographic data comprises sales date, sales area, sales location, dealer and channel classification; the product background data comprises a product serial number, a product category and a product price;
and a composition module: obtaining a sales graph according to the sales geographic data; five sales background data are correspondingly obtained to obtain five sales figures, and the sales background data are in one-to-one correspondence with the sales figures; the sales map is a two-dimensional image marked with sales background data;
and a product coefficient module: obtaining a product coefficient according to the product serial number and the product sales relationship;
and a seasonal coefficient module: according to the seasons, obtaining seasonal coefficients through a seasonal sales relationship;
and a prediction module: obtaining a plurality of sales volumes through a sales relationship network according to the product coefficients, the five sales figures and the seasonal coefficients; one product category corresponds to one sales volume;
the sales relationship network comprises a two-dimensional convolution network, a three-dimensional convolution network and a fusion network; the converged network comprises a first converged network, a second converged network and a third converged network.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the method of any one of claims 1-8 when the computer program is executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310911626.7A CN116911908B (en) | 2023-07-25 | 2023-07-25 | Sales data prediction method and system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310911626.7A CN116911908B (en) | 2023-07-25 | 2023-07-25 | Sales data prediction method and system based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116911908A true CN116911908A (en) | 2023-10-20 |
| CN116911908B CN116911908B (en) | 2024-02-27 |
Family
ID=88357899
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310911626.7A Active CN116911908B (en) | 2023-07-25 | 2023-07-25 | Sales data prediction method and system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116911908B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113256007A (en) * | 2021-05-28 | 2021-08-13 | 浙江大学 | Multi-mode-oriented new product sales forecasting method and device |
| CN114170162A (en) * | 2021-11-25 | 2022-03-11 | 深圳先进技术研究院 | Image prediction method, image prediction device and computer storage medium |
| CN114282657A (en) * | 2021-12-30 | 2022-04-05 | 浙江中控技术股份有限公司 | Market data long-term prediction model training method, device, equipment and storage medium |
| US20220261928A1 (en) * | 2019-07-08 | 2022-08-18 | Indigo Ag, Inc. | Crop yield forecasting models |
| CN115880027A (en) * | 2022-12-12 | 2023-03-31 | 深圳市宝视佳科技有限公司 | Electronic commerce website commodity seasonal prediction model creation method |
-
2023
- 2023-07-25 CN CN202310911626.7A patent/CN116911908B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20220261928A1 (en) * | 2019-07-08 | 2022-08-18 | Indigo Ag, Inc. | Crop yield forecasting models |
| CN113256007A (en) * | 2021-05-28 | 2021-08-13 | 浙江大学 | Multi-mode-oriented new product sales forecasting method and device |
| CN114170162A (en) * | 2021-11-25 | 2022-03-11 | 深圳先进技术研究院 | Image prediction method, image prediction device and computer storage medium |
| CN114282657A (en) * | 2021-12-30 | 2022-04-05 | 浙江中控技术股份有限公司 | Market data long-term prediction model training method, device, equipment and storage medium |
| CN115880027A (en) * | 2022-12-12 | 2023-03-31 | 深圳市宝视佳科技有限公司 | Electronic commerce website commodity seasonal prediction model creation method |
Non-Patent Citations (1)
| Title |
|---|
| 陈加;张玉麒;宋鹏;魏艳涛;王煜;: "深度学习在基于单幅图像的物体三维重建中的应用", 自动化学报, no. 04 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116911908B (en) | 2024-02-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110415516B (en) | Urban traffic flow prediction method and medium based on graph convolution neural network | |
| CN107330731B (en) | Method and device for identifying click abnormity of advertisement space | |
| CN110579359B (en) | Optimization method and system of automatic driving failure scene library, server and medium | |
| CN111680701B (en) | Training method and device of image recognition model and image recognition method and device | |
| CN107688772A (en) | Method, apparatus, computer equipment and the storage medium of policy information typing | |
| CN113989305B (en) | Target semantic segmentation method and street target abnormity detection method applying same | |
| CN110728295B (en) | Semi-supervised landform classification model training and landform graph construction method | |
| Sakieh et al. | Performance assessment of geospatial simulation models of land-use change—a landscape metric-based approach | |
| CN117274608B (en) | Remote sensing image semantic segmentation method based on space detail perception and attention guidance | |
| CN111414878B (en) | Social attribute analysis and image processing method and device for land parcels | |
| CN111881243A (en) | Taxi track hotspot area analysis method and system | |
| CN115512258A (en) | Desensitization method and device for video image, terminal equipment and storage medium | |
| CN114612402A (en) | Method, device, equipment, medium and program product for determining object quantity | |
| CN113222209B (en) | Regional tail gas migration prediction method and system based on domain adaptation and storage medium | |
| CN116911908B (en) | Sales data prediction method and system based on artificial intelligence | |
| CN114820922A (en) | Automatic driving effective static scene construction method and system based on complexity | |
| CN114359293A (en) | Three-dimensional MRI brain tumor segmentation method based on deep learning | |
| CN111126561B (en) | Image processing method based on multi-path parallel convolution neural network | |
| CN111121803A (en) | Method and device for acquiring common stop points of road | |
| CN114782821B (en) | Coastal wetland vegetation remote sensing identification method combined with multiple migration learning strategies | |
| CN115497075A (en) | Traffic target detection method based on improved convolutional neural network and related device | |
| CN115049836A (en) | Image segmentation method, device, equipment and storage medium | |
| Gupta et al. | Risks in supply chain management and its mitigation | |
| CN117541883B (en) | Image generation model training, image generation method, system and electronic equipment | |
| CN113807428B (en) | Reconstruction method, system and device of classification model probability label and storage medium |
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 |