CN116245666A - Cost accounting method and system based on data processing - Google Patents
Cost accounting method and system based on data processing Download PDFInfo
- Publication number
- CN116245666A CN116245666A CN202310056423.4A CN202310056423A CN116245666A CN 116245666 A CN116245666 A CN 116245666A CN 202310056423 A CN202310056423 A CN 202310056423A CN 116245666 A CN116245666 A CN 116245666A
- Authority
- CN
- China
- Prior art keywords
- data
- expense
- sample
- picture
- real
- 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
Images
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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/12—Accounting
- G06Q40/125—Finance or payroll
-
- 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
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- 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)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Finance (AREA)
- Software Systems (AREA)
- Accounting & Taxation (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Technology Law (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention provides a cost accounting method and a system based on data processing, which comprises the steps of classifying and segmenting historical cost data and marking the historical cost data in a coordinate system, converting the data in the coordinate system into sample pictures through a sliding window, training a classification model through periodic sample pictures, amplifying or shrinking normal data to obtain negative samples, training a discrimination network through the positive and negative samples, and judging whether the current cost data is abnormal through the trained discrimination model. The technical problem that whether the stable data are normal or not can only be judged in the prior art is solved through the scheme.
Description
Technical Field
The invention relates to the field of financial data processing, in particular to a cost accounting method and system based on data processing.
Background
In the financial management activities of enterprises, various kinds of fees of the enterprises need to be accounted for, and in the prior art, the accounting of the fees is usually carried out manually or automatically through a computer. The automatic processing of the computer is better for the calculated value, such as the calculation of total cost, the classification of total cost, etc., but the identification of the financial abnormal data is more difficult, especially the non-fixed cost, such as the electricity cost, water cost, the periodically changing product raw material cost, etc. which are changed with seasons, and the real-time identification is difficult if the cost is abnormal. In addition, prior art fee accounting systems are typically at the end of the financial cycle, such as the end of the month, the end of the quarter, etc., resulting in heavy financial staff effort during the fee accounting period.
Disclosure of Invention
In order to solve the problems in the background art, the invention provides a cost accounting method and a cost accounting system based on data processing, which adopt a neural network model to identify abnormal points in periodic data in real time and reduce the working pressure of financial staff.
In one aspect of the present invention, there is provided a cost accounting method based on data processing, characterized in that the method comprises the steps of: acquiring historical cost data with no accounting errors, and classifying the historical cost data according to service types to obtain original cost data; combining the original expense data according to a first preset time length to obtain segmented expense data, wherein the segmented expense data is time sequence data; representing the segmented cost data on an XY coordinate system, wherein X is time, Y is combined cost within the first preset time length, and obtaining a segmented cost data coordinate system; taking a second preset time length as a moving window to carry out moving interception on the segmented expense data coordinate system to obtain sample expense data; converting all the sample expense data into pictures with the same size to obtain a first sample picture; classifying the first sample picture into a periodic sample picture and an aperiodic sample picture according to whether the data points in the first sample picture have periodicity; training a neural network model by using the periodic sample pictures and the aperiodic sample pictures to obtain a first model, wherein the first model can classify the periodic data and the aperiodic data of the input pictures; selecting a preset number of sample expense data, selecting at least one data point for each sample expense data, and amplifying or reducing the data of the selected data points to obtain negative sample expense data; converting the negative sample expense data into pictures with consistent sizes to obtain a negative sample picture; training a neural network model by using the periodic sample picture and the negative sample picture to obtain a second model, wherein the second model can identify whether an abnormal charge data point exists in the input picture; acquiring real-time expense data, and merging the real-time expense data according to a first preset time length to obtain real-time segmented expense data; representing the real-time segmentation expense data on an XY coordinate system to obtain a real-time segmentation expense data coordinate system; intercepting the real-time segmented expense data coordinate system by taking the latest data as a starting point and the second preset time length to obtain real-time expense data; converting the real-time expense data into a picture to obtain a picture to be detected; and inputting the picture to be detected into a first model, and if the picture to be detected is judged to be periodic data, inputting the picture to be detected into a second model, and outputting whether the picture to be detected has abnormal charge data points by the second model.
Further, the first preset time lengths corresponding to different services are different.
Further, the second preset time lengths corresponding to different services are different.
Further, the second preset time length is greater than the first preset time length.
Further, the magnification exceeds the second threshold multiple, or the reduction is less than a third threshold fraction.
In another aspect, the present invention also provides a fee accounting system based on data processing, which is characterized in that the system includes the following modules: the classification module is used for acquiring historical cost data with no accounting errors, classifying the historical cost data according to the service type and obtaining original cost data; the segmentation module is used for merging the original expense data according to a first preset time length to obtain segmentation expense data, wherein the segmentation expense data is time sequence data; the representation module is used for representing the segmentation expense data on an XY coordinate system, wherein X is time, Y is combined expense in the first preset time length, and a segmentation expense data coordinate system is obtained; the intercepting module is used for movably intercepting the segmented expense data coordinate system by taking a second preset time length as a moving window to obtain sample expense data; the conversion module is used for converting all the sample expense data into pictures with the same size to obtain a first sample picture; the labeling module is used for classifying the first sample picture into a periodic sample picture and an aperiodic sample picture according to whether the data points in the first sample picture have periodicity; the first training module is used for training the neural network model by using the periodic sample pictures and the aperiodic sample pictures to obtain a first model, and the first model can classify the periodic data and the aperiodic data of the input pictures; the negative sample module is used for selecting a preset number of sample expense data, selecting at least one data point for each sample expense data, and amplifying or reducing the data of the selected data points to obtain negative sample expense data; the second conversion module is used for converting the negative sample expense data into pictures with the same size to obtain a negative sample picture; the second training module is used for training the neural network model by using the periodic sample picture and the negative sample picture to obtain a second model, and the second model can identify whether an abnormal charge data point exists in the input picture; the real-time data module is used for acquiring real-time expense data, and combining the real-time expense data according to a first preset time length to obtain real-time segmented expense data; the second segmentation module is used for representing the real-time segmentation expense data on an XY coordinate system to obtain a real-time segmentation expense data coordinate system; the second intercepting module is used for intercepting the real-time segmented expense data coordinate system by taking the latest data as a starting point and the second preset time length to obtain real-time expense data; the detection module is used for converting the real-time expense data into pictures to obtain pictures to be detected; and inputting the picture to be detected into a first model, and if the picture to be detected is judged to be periodic data, inputting the picture to be detected into a second model, and outputting whether the picture to be detected has abnormal charge data points by the second model.
According to the technical scheme, abnormal points in periodic data are identified in real time by adopting the neural network model, so that the working pressure of financial staff is reduced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a diagram of a cost data coordinate system;
FIG. 2 is a schematic diagram of a sliding window;
FIG. 3 is a sample cost data schematic;
FIG. 4 is a first sample picture schematic;
fig. 5 is a schematic diagram of real-time data interception.
Description of the embodiments
The invention will be described with reference to the drawings and detailed description.
The present embodiment solves the above problem by:
in one embodiment, the invention discloses a cost accounting method based on data processing, which comprises the following steps:
and acquiring historical cost data with no accounting errors, and classifying the historical cost data according to the service type to obtain original cost data.
The historical cost data may be data within a financial system of an enterprise or organization, with the historical cost data being as much as possible for the accuracy of the model. Due to the different cost items, the cost can be divided into multiple types, such as rentals, hydropower, different raw material costs, employee payouts, etc., and the different types of costs can fluctuate differently; the house renting cost is usually stable, and for stable data, the existing accounting system can easily judge whether abnormal cost exists or not, and real-time early warning can be carried out only by setting a certain fluctuation threshold; however, most of the data are not stable, such as electricity charge data, the electricity charge in working days is higher than that in rest days, the price of source materials in different seasons is different, the raw material charge fluctuates seasonally, and more false alarms can be caused if a threshold value is set for the fluctuation charge; also, some data are completely irregular, such as short-term business trip cost, and the business trip cost may randomly fluctuate due to the randomness of business trip business; for stable data, a better solution exists in the prior art, and for irregular data, effective check cannot be performed, and the method is mainly used for classifying the data according to the regularly fluctuating data, so that the fluctuation type of the data can be found out; raw cost data is obtained by classifying the data, the raw cost data may include long time cost data, and each cost data includes a value of a cost and a time point of occurrence. The existing financial system generally stores historical cost data with no accounting errors, and the historical cost data with no accounting errors is more convenient to obtain, so that the invention only obtains the historical cost data with no accounting errors.
And merging the original expense data according to a first preset time length to obtain segmented expense data, wherein the segmented expense data is time sequence data.
For convenience of periodical statistics, merging the original expense data according to a first preset time length, wherein merging refers to overlapping expense data belonging to the same time period; if Zhou Wei the first preset time length is adopted, the cost in the first week is added, the cost in the second week is added, and the like, and all the historical data are combined and overlapped according to the main cycle of the week to obtain the segmented cost data segmented by the week. Since the segment cost data are sequentially combined according to the preset time length, all the combined data form a time sequence.
The first preset time length corresponding to different businesses is different, for example, the water charge and electricity charge can be used as a period of days, the employee wage expense can be used as a period of months, the raw material charge can be used as a period of quarters, and specific selection needs to be determined according to the charge fluctuation condition of different businesses and the data density degree.
For the missing data, the missing data may be set to the data of the previous cycle, for example, the electricity charge of the first week is a-ary, the electricity charge of the third week is b-ary, and the electricity charge of the second week is a-ary.
And the segmentation expense data is expressed on an XY coordinate system, wherein X is time, Y is the combined expense in the first preset time length, and the segmentation expense data coordinate system is obtained.
Since the segment cost data is time-series data, the data is related to time and cost, and for convenience of image recognition, all segment cost data are represented in an XY coordinate system, and as shown in fig. 1, all data are marked on the coordinate system.
And taking the second preset time length as a moving window to carry out moving interception on the segmented expense data coordinate system, so as to obtain sample expense data.
Since very much data is needed for model training, and the reference data of the same type of service is organized together, in order to obtain enough samples, a sliding window is used for sliding and intercepting the samples. As shown in fig. 2, the short dashed line is an example of a sliding window, and after the data is intercepted, the sliding window slides backward to the long dashed line (in fig. 2, the two windows are staggered for convenience in distinguishing the sliding window, and in actual operation, the two windows are partially overlapped).
As shown in fig. 3, a part of the data is cut out by the sliding window.
Since the periodicity of the different services may be different, the second preset time length of the different services may be set to be different; further, in order to preserve the periodicity of the intercepted data, if the service data is periodic data, the second preset time period should cover at least two periods, so that the model can identify the periodicity of the intercepted data in the subsequent steps. In addition, the sliding window slides for a third preset time length, as shown in fig. 2, the first sliding window is cut from the starting point, the second sliding window is cut from the 6 th time period, and the sliding distance parameter of the sliding window is 6 first time periods; the specific sliding distance parameter can be set according to the data, and although the smaller the sliding distance parameter is, the more data is intercepted, the more samples are.
In addition, since the sliding window needs to intercept data of a plurality of first time lengths, the second preset time length is, though, greater than the first preset time length.
And converting all the sample expense data into pictures with the same size, and obtaining a first sample picture.
Because the data density of different services is different and the periodicity is different, if discrete data is adopted to train a machine learning model, the number of undetermined parameters is excessive, and the model is very difficult to converge; meanwhile, because the image classification model in the existing machine learning model is more in research, the model which can be referred is more, and the model training is faster than the discrete data training, the invention converts the sample cost data into the image, and the image is used as the sample for training the model. In addition, in order to facilitate the input of the model, when the image conversion is performed, all sample expense data are converted into pictures with the same size, as shown in fig. 4, the intercepted data can be directly scaled to achieve the same size.
Classifying the first sample picture into a periodic sample picture and an aperiodic sample picture according to whether the data points in the first sample picture have periodicity.
As described above, the present invention is mainly directed to a core of periodic data, so that it is necessary to distinguish periodic data from non-periodic data, and different labels are applied to different data, so that the periodic data can be regarded as a positive sample, and the non-periodic data can be regarded as a negative sample. The image classification can be performed manually or by any other automatic method, and the invention is not limited.
Training a neural network model by using the periodic sample pictures and the aperiodic sample pictures to obtain a first model, wherein the first model can classify the periodic data and the aperiodic data of the input pictures;
after a sample with enough sample is obtained, the model can be trained, and because the data points are converted into images, the neural network model can adopt any mature model in the prior art, such as VGG model, inceptionV2 and the like. Periodic and aperiodic data are used for identification, so that the trained first model can classify the periodic data and the aperiodic data of the input picture.
Because the principle of the first model is pattern recognition, the first model can recognize the periodic waveform in the image only, so that the first model can be trained, new business can be used for classifying new types of data without manual periodic labeling after the first model is trained, manual work can be greatly reduced when the new business exists, and in addition, the first model can be integrated into a system after the first model is trained and is transferred to other users for use.
Selecting a preset number of sample expense data, selecting at least one data point for each sample expense data, and amplifying or reducing the data of the selected data points to obtain negative sample expense data.
Since the raw data are all historical cost data with no verification errors, the samples are positive samples with no verification errors, and the model training usually needs negative samples. Because the erroneous cost data deviate from the normal data, at least one data point is arbitrarily selected from the sample cost data to be amplified or reduced to a certain extent, so that an abnormal data point can be obtained, and if the training model can identify the data point, abnormal cost can be identified. Further, in order to reduce the error, the degree of enlargement or reduction should reach a predetermined value, such as enlarging a predetermined number of bits, or reducing to a fraction of the original predetermined number.
Converting the negative sample expense data into pictures with consistent sizes to obtain a negative sample picture;
similar to the reason for converting positive sample data into pictures, negative sample cost data also requires a picture conversion, and similarly, the acquired pictures are scaled to obtain a size suitable for model input, and the obtained pictures are images containing abnormal cost, namely, negative sample pictures.
And training a neural network model by using the periodic sample picture and the negative sample picture to obtain a second model, wherein the second model can identify whether an abnormal charge data point exists in the input picture.
After the positive and negative samples of certain data are obtained, the training of the second neural network lever can be performed, and the second neural network model can select an countermeasure model, such as a GAN network or any improvement on the GAN network in the prior art, and of course, other models capable of being trained by using the positive and negative samples can also be selected. The second model can identify whether the input picture has abnormal charge data points, and when the abnormal charge data points are identified, the user can be reminded of abnormal charge data, and the user is prompted to check manually, so that the manual workload is reduced.
And acquiring real-time expense data, and merging the real-time expense data according to a first preset time length to obtain real-time segmented expense data.
In order to reduce the task of not checking in the charge period, the invention acquires the real-time charge data, and the invention needs to say that the acquired real-time charge data comprises data with a period of time, the specific period of time can be set differently according to the service type and the service periodicity, such as the electricity charge, which is usually periodic in a week, so the acquired real-time charge data can be data of a month; payroll is periodic in months and acquiring real-time cost data may be data of the last year. Similar to the operation during training, the real-time data are combined to obtain real-time segmented cost data, i.e. a piece of data to be processed currently.
And the real-time segmentation expense data is expressed on an XY coordinate system to obtain a real-time segmentation expense data coordinate system.
In the identification, the data is also required to be converted into an image, so that the real-time segmentation expense data is also required to be expressed on an XY coordinate system, and the real-time segmentation expense data coordinate system is obtained.
And intercepting the real-time segmented expense data coordinate system by taking the latest data as a starting point and the second preset time length to obtain real-time expense data.
In the cost accounting, the most recent cost is calculated, so that the intercepted data is also the most recent data, and in order to make the scale of the image consistent with that in training, the intercepted window also adopts a second preset time length. As shown in fig. 5, when real-time data is intercepted, only the rightmost window data needs to be intercepted, so that when the data is identified, whether the data in the rightmost window is abnormal or not can be identified.
Converting the real-time expense data into a picture to obtain a picture to be detected; and inputting the picture to be detected into a first model, and if the picture to be detected is judged to be periodic data, inputting the picture to be detected into a second model, and outputting whether the picture to be detected has abnormal charge data points by the second model.
Similar to the training situation, the relevant data can be converted into images after the real-time expense data is obtained, and because new services can exist during the running period of the system, each image still needs to be periodically judged, and only after the periodic data is judged, the data form requirement of the system is met, so that the next judgment can be performed. When the second model judges that the cost data in the image has abnormal points, the user can be informed that the data of the window is abnormal and further accounting needs to be carried out manually.
In another embodiment, the invention also discloses a cost accounting system based on data processing, which is characterized by comprising the following modules:
the classification module is used for acquiring historical cost data with no accounting errors, classifying the historical cost data according to the service type and obtaining original cost data;
the segmentation module is used for merging the original expense data according to a first preset time length to obtain segmentation expense data, wherein the segmentation expense data is time sequence data;
the representation module is used for representing the segmentation expense data on an XY coordinate system, wherein X is time, Y is combined expense in the first preset time length, and a segmentation expense data coordinate system is obtained;
the intercepting module is used for movably intercepting the segmented expense data coordinate system by taking a second preset time length as a moving window to obtain sample expense data;
the conversion module is used for converting all the sample expense data into pictures with the same size to obtain a first sample picture;
the labeling module is used for classifying the first sample picture into a periodic sample picture and an aperiodic sample picture according to whether the data points in the first sample picture have periodicity;
the first training module is used for training the neural network model by using the periodic sample pictures and the aperiodic sample pictures to obtain a first model, and the first model can classify the periodic data and the aperiodic data of the input pictures;
the negative sample module is used for selecting a preset number of sample expense data, selecting at least one data point for each sample expense data, and amplifying or reducing the data of the selected data points to obtain negative sample expense data;
the second conversion module is used for converting the negative sample expense data into pictures with the same size to obtain a negative sample picture;
the second training module is used for training the neural network model by using the periodic sample picture and the negative sample picture to obtain a second model, and the second model can identify whether an abnormal charge data point exists in the input picture;
the real-time data module is used for acquiring real-time expense data, and combining the real-time expense data according to a first preset time length to obtain real-time segmented expense data;
the second segmentation module is used for representing the real-time segmentation expense data on an XY coordinate system to obtain a real-time segmentation expense data coordinate system;
the second intercepting module is used for intercepting the real-time segmented expense data coordinate system by taking the latest data as a starting point and the second preset time length to obtain real-time expense data;
the detection module is used for converting the real-time expense data into pictures to obtain pictures to be detected; and inputting the picture to be detected into a first model, and if the picture to be detected is judged to be periodic data, inputting the picture to be detected into a second model, and outputting whether the picture to be detected has abnormal charge data points by the second model.
It should be noted that the detailed implementation principle and further improvement measures of the above-mentioned fee accounting system based on data processing are the same as those of the foregoing fee accounting method based on data processing, and will not be described in detail in this embodiment, and those skilled in the art may implement the detailed implementation in the system according to the existing fee accounting method based on data processing.
Finally, it should be noted that: the above embodiments are only for illustrating the technical aspects of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: modifications and equivalents may be made to the specific embodiments of the invention without departing from the spirit and scope of the invention, which is intended to be covered by the claims.
The present invention is not limited to the specific partial module structure described in the prior art. The prior art to which this invention refers in the preceding background section as well as in the detailed description section can be used as part of the invention for understanding the meaning of some technical features or parameters. The protection scope of the present invention is subject to what is actually described in the claims.
Claims (10)
1. A method of cost accounting based on data processing, the method comprising the steps of:
acquiring historical cost data with no accounting errors, and classifying the historical cost data according to service types to obtain original cost data;
combining the original expense data according to a first preset time length to obtain segmented expense data, wherein the segmented expense data is time sequence data;
representing the segmented cost data on an XY coordinate system, wherein X is time, Y is combined cost within the first preset time length, and obtaining a segmented cost data coordinate system;
taking a second preset time length as a moving window to carry out moving interception on the segmented expense data coordinate system to obtain sample expense data;
converting all the sample expense data into pictures with the same size to obtain a first sample picture;
classifying the first sample picture into a periodic sample picture and an aperiodic sample picture according to whether the data points in the first sample picture have periodicity;
training a neural network model by using the periodic sample pictures and the aperiodic sample pictures to obtain a first model, wherein the first model can classify the periodic data and the aperiodic data of the input pictures;
selecting a preset number of sample expense data, selecting at least one data point for each sample expense data, and amplifying or reducing the data of the selected data points to obtain negative sample expense data;
converting the negative sample expense data into pictures with consistent sizes to obtain a negative sample picture;
training a neural network model by using the periodic sample picture and the negative sample picture to obtain a second model, wherein the second model can identify whether an abnormal charge data point exists in the input picture;
acquiring real-time expense data, and merging the real-time expense data according to a first preset time length to obtain real-time segmented expense data;
representing the real-time segmentation expense data on an XY coordinate system to obtain a real-time segmentation expense data coordinate system;
intercepting the real-time segmented expense data coordinate system by taking the latest data as a starting point and the second preset time length to obtain real-time expense data;
converting the real-time expense data into a picture to obtain a picture to be detected; and inputting the picture to be detected into a first model, and if the picture to be detected is judged to be periodic data, inputting the picture to be detected into a second model, and outputting whether the picture to be detected has abnormal charge data points by the second model.
2. The data processing-based fee accounting method according to claim 1, wherein: the first preset time lengths corresponding to different services are different.
3. The data processing-based fee accounting method according to claim 2, wherein: the second preset time lengths corresponding to different services are different.
4. A cost accounting method based on data processing according to claim 3, characterized in that: the second preset time length is greater than the first preset time length.
5. The data processing-based fee accounting method according to claim 1, wherein: zooming in or out the data of the selected data point includes: the magnification exceeds the second threshold multiple, or the demagnification is less than a third threshold fraction.
6. A cost accounting system based on data processing, characterized in that the system comprises the following modules:
the classification module is used for acquiring historical cost data with no accounting errors, classifying the historical cost data according to the service type and obtaining original cost data;
the segmentation module is used for merging the original expense data according to a first preset time length to obtain segmentation expense data, wherein the segmentation expense data is time sequence data;
the representation module is used for representing the segmentation expense data on an XY coordinate system, wherein X is time, Y is combined expense in the first preset time length, and a segmentation expense data coordinate system is obtained;
the intercepting module is used for movably intercepting the segmented expense data coordinate system by taking a second preset time length as a moving window to obtain sample expense data;
the conversion module is used for converting all the sample expense data into pictures with the same size to obtain a first sample picture;
the labeling module is used for classifying the first sample picture into a periodic sample picture and an aperiodic sample picture according to whether the data points in the first sample picture have periodicity;
the first training module is used for training the neural network model by using the periodic sample pictures and the aperiodic sample pictures to obtain a first model, and the first model can classify the periodic data and the aperiodic data of the input pictures;
the negative sample module is used for selecting a preset number of sample expense data, selecting at least one data point for each sample expense data, and amplifying or reducing the data of the selected data points to obtain negative sample expense data;
the second conversion module is used for converting the negative sample expense data into pictures with the same size to obtain a negative sample picture;
the second training module is used for training the neural network model by using the periodic sample picture and the negative sample picture to obtain a second model, and the second model can identify whether an abnormal charge data point exists in the input picture;
the real-time data module is used for acquiring real-time expense data, and combining the real-time expense data according to a first preset time length to obtain real-time segmented expense data;
the second segmentation module is used for representing the real-time segmentation expense data on an XY coordinate system to obtain a real-time segmentation expense data coordinate system;
the second intercepting module is used for intercepting the real-time segmented expense data coordinate system by taking the latest data as a starting point and the second preset time length to obtain real-time expense data;
the detection module is used for converting the real-time expense data into pictures to obtain pictures to be detected; and inputting the picture to be detected into a first model, and if the picture to be detected is judged to be periodic data, inputting the picture to be detected into a second model, and outputting whether the picture to be detected has abnormal charge data points by the second model.
7. The data processing based fee accounting system of claim 6 wherein: the first preset time lengths corresponding to different services are different.
8. The data processing based fee accounting system of claim 7, wherein: the second preset time lengths corresponding to different services are different.
9. The data processing based fee accounting system of claim 8, wherein: the second preset time length is greater than the first preset time length.
10. The data processing based fee accounting system of claim 6 wherein: zooming in or out the data of the selected data point includes: the magnification exceeds the second threshold multiple, or the demagnification is less than a third threshold fraction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310056423.4A CN116245666B (en) | 2023-01-16 | 2023-01-16 | Cost accounting method and system based on data processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310056423.4A CN116245666B (en) | 2023-01-16 | 2023-01-16 | Cost accounting method and system based on data processing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116245666A true CN116245666A (en) | 2023-06-09 |
| CN116245666B CN116245666B (en) | 2023-09-19 |
Family
ID=86630686
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310056423.4A Active CN116245666B (en) | 2023-01-16 | 2023-01-16 | Cost accounting method and system based on data processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116245666B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108230131A (en) * | 2017-12-29 | 2018-06-29 | 国信优易数据有限公司 | A kind of data processing method and device |
| CN112861906A (en) * | 2020-12-31 | 2021-05-28 | 深圳前海微众银行股份有限公司 | Method and device for detecting picture abnormity |
| CN114154866A (en) * | 2021-12-02 | 2022-03-08 | 北京顶象技术有限公司 | Marketing enterprise financial risk early warning method and system |
| CN114881123A (en) * | 2022-04-21 | 2022-08-09 | 南开大学 | Method and device for detecting periodic data abnormity, electronic equipment and storage medium |
| CN115271929A (en) * | 2022-08-17 | 2022-11-01 | 深圳前海微众银行股份有限公司 | Data anomaly detection method, device, equipment, medium and program product |
| US11494850B1 (en) * | 2019-03-13 | 2022-11-08 | Alight Solutions Llc | Applied artificial intelligence technology for detecting anomalies in payroll data |
| CN115482075A (en) * | 2022-09-21 | 2022-12-16 | 平安银行股份有限公司 | Financial data anomaly analysis method and device, electronic equipment and storage medium |
-
2023
- 2023-01-16 CN CN202310056423.4A patent/CN116245666B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108230131A (en) * | 2017-12-29 | 2018-06-29 | 国信优易数据有限公司 | A kind of data processing method and device |
| US11494850B1 (en) * | 2019-03-13 | 2022-11-08 | Alight Solutions Llc | Applied artificial intelligence technology for detecting anomalies in payroll data |
| CN112861906A (en) * | 2020-12-31 | 2021-05-28 | 深圳前海微众银行股份有限公司 | Method and device for detecting picture abnormity |
| CN114154866A (en) * | 2021-12-02 | 2022-03-08 | 北京顶象技术有限公司 | Marketing enterprise financial risk early warning method and system |
| CN114881123A (en) * | 2022-04-21 | 2022-08-09 | 南开大学 | Method and device for detecting periodic data abnormity, electronic equipment and storage medium |
| CN115271929A (en) * | 2022-08-17 | 2022-11-01 | 深圳前海微众银行股份有限公司 | Data anomaly detection method, device, equipment, medium and program product |
| CN115482075A (en) * | 2022-09-21 | 2022-12-16 | 平安银行股份有限公司 | Financial data anomaly analysis method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116245666B (en) | 2023-09-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021052031A1 (en) | Statistical interquartile range-based commodity inventory risk early warning method and system, and computer readable storage medium | |
| CN110472524B (en) | Invoice information management method and system based on deep learning and readable medium | |
| CN111401314A (en) | Dressing information detection method, device, equipment and storage medium | |
| CN113762224B (en) | Engineering cost achievement quality inspection system and method thereof | |
| CN104702492A (en) | Garbage message model training method, garbage message identifying method and device thereof | |
| WO2020062702A1 (en) | Method and device for sending text messages, computer device and storage medium | |
| CN111786385B (en) | Power grid operation and maintenance scheme planning method, system and equipment | |
| CN110634081A (en) | Method and device for processing abnormal data of hydropower station | |
| CN115277180B (en) | Block chain log anomaly detection and tracing system | |
| CN114648393A (en) | Data mining method, system and equipment applied to bidding | |
| CN117473048B (en) | Financial abnormal data monitoring and analyzing system and method based on data mining | |
| CN117035419B (en) | Intelligent management system and method for enterprise project implementation | |
| CN115423586A (en) | Financial invoice reimbursement, uploading and auditing system based on network | |
| CN112150098A (en) | Electric power facility anti-terrorism security supervision method and system | |
| CN116245666B (en) | Cost accounting method and system based on data processing | |
| CN109522309A (en) | One kind being based on auditing system procurement information recording exceptional value processing method | |
| CN109801394B (en) | Staff attendance checking method and device, electronic equipment and readable storage medium | |
| Vorozhtsova et al. | System approach to development of intellectual information mobile system for electric power metering | |
| CN115660475A (en) | System and method for managing work efficiency of function department of smart card manufacturing enterprise | |
| CN115482075A (en) | Financial data anomaly analysis method and device, electronic equipment and storage medium | |
| CN115471794A (en) | Intelligent power plant small target detection method and system based on YOLOv5 | |
| CN114039279A (en) | Control cabinet monitoring method and system in rail transit station | |
| CN109976967B (en) | Payment and recovery monitoring and early warning method and system based on intelligent scheduling | |
| CN111612302A (en) | Group-level data management method and equipment | |
| CN111506626B (en) | Pass gate control system based on ship navigation behavior information analysis |
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 |