An enterprise attendance management system based on big data analysis

Abstract

The invention relates to the technical field of big data analysis, in particular to an enterprise attendance management system based on big data analysis, which comprises a data collection module, a data analysis module and an attendance management module, wherein: the data collection module collects employee attendance records; the data analysis module is used to analyze the employee attendance records, and build an isolation forest algorithm model according to the employee attendance records, wherein the isolation forest algorithm model includes a plurality of isolated trees; the shortest path length and the average path length of each employee data in all isolated trees are calculated; the abnormality score of each data point is calculated according to the shortest path length and the average path length; whether the employee attendance is abnormal is determined according to the size of the abnormality score, and the abnormal features in the employee attendance record are determined according to the shortest path length; the attendance management module is used to generate an abnormal attendance report according to the abnormal features in the employee attendance record, and provide management suggestions.

Description

Technical Field

[0001] The present invention relates to the technical field of big data analysis, and in particular to an enterprise attendance management system based on big data analysis. Background Art

[0002] Traditional enterprise attendance management systems are often managed through processes such as data collection, data processing, data analysis, and user interaction interfaces. Data analysis often uses big data analysis technologies, such as isolation forests, support vector machines, and other algorithms, to conduct in-depth analysis of attendance data, as follows:

[0003] When calculating the anomaly score, the traditional isolation forest algorithm usually only uses the average path length of the data point in all isolated trees. Although this method can capture the overall abnormality of the data point, it ignores the extreme abnormal characteristics that some data points may exhibit in some isolated trees. This leads to insufficient accuracy and comprehensiveness of anomaly detection;

[0004] Moreover, after determining the abnormal data points, they often only provide simple abnormal labels, without further analyzing the specific characteristics and causes of the abnormalities. This generalized abnormal labeling method makes it difficult for management to take targeted management measures, reducing management effectiveness. Summary of the invention

[0005] The purpose of the present invention is to provide an enterprise attendance management system based on big data analysis to solve the problems raised in the above background technology.

[0006] To achieve the above purpose, the present invention provides the following technical solution: an enterprise attendance management system based on big data analysis, comprising a data collection module, a data analysis module and an attendance management module, wherein:

[0007] The data collection module collects employee attendance records;

[0008] The data analysis module is used to analyze employee attendance records, and build an isolation forest algorithm model based on the employee attendance records, wherein the isolation forest algorithm model includes multiple isolated trees; calculate the shortest path length and average path length of each employee data in all isolated trees; calculate the abnormality score of each data point based on the shortest path length and the average path length; determine whether the employee attendance is abnormal based on the size of the abnormality score, and determine the abnormal characteristics in the employee attendance record based on the shortest path length;

[0009] The attendance management module is used to generate abnormal attendance reports based on abnormal features in employee attendance records and provide management suggestions.

[0010] As a further improvement of the technical solution, the data collection module collects employee attendance records from different data sources, wherein the employee attendance records include multiple attendance features.

[0011] As a further improvement of the technical solution, the data analysis module includes a model building unit, and the process of the model building unit building the isolation forest algorithm model specifically includes:

[0012] Randomly select an attendance feature from different attendance features in the employee attendance record, and randomly select a split point within the value range of the selected attendance feature;

[0013] The data set is divided into two subsets according to the split point, one subset contains data points with eigenvalues ​​less than the split point, and the other subset contains data points with eigenvalues ​​greater than or equal to the split point;

[0014] Recursively perform the above steps for each subset until the number of data points in the subset is less than the preset threshold or the maximum tree depth is reached;

[0015] According to the above steps, multiple isolated trees are repeatedly constructed to form an isolation forest.

[0016] As a further improvement of the technical solution, the data analysis module includes a path analysis unit, and the process of calculating the path length by the path analysis unit specifically includes:

[0017] For each isolated tree in the isolation forest, input each employee attendance record data point;

[0018] Starting from the root node, recursively assign data points to corresponding child nodes according to features and split points until the data points reach the leaf node;

[0019] Record the path length of the data point from the root node to the leaf node;

[0020] The path length of each employee attendance record data point in all isolated trees is stored in a matrix, where the rows of the matrix correspond to the data points and the columns correspond to the isolated trees.

[0021] As a further improvement of the technical solution, the path analysis unit calculates the shortest path length by finding the minimum value from the corresponding row of the path length matrix to determine the shortest path of each employee attendance record data point in all isolated trees.

[0022] As a further improvement of the present technical solution, the path analysis unit calculates the average path length by calculating the average value of all path lengths from the corresponding rows of the path length matrix, by summing the extracted path lengths and then dividing by the number of isolated trees.

[0023] As a further improvement of the technical solution, the data analysis module includes an abnormality score calculation unit, and the abnormality score calculation process of the abnormality score calculation unit specifically includes:

[0024] The shortest path length and the average path length are combined to form a comprehensive path length, which is calculated using the weighted sum method. The formula is as follows:

[0025] ,in represents the comprehensive path length, represents the shortest path length; represents the average path length; and Represent the weights of the shortest path length and the average path length respectively;

[0026] The formula for calculating the abnormal score of each employee's attendance record data point is as follows:

[0027] ,in represents the abnormality score, It is a constant representing the expected path length of an isolated tree, which is ;

[0028] in, is the number of data points.

[0029] As a further improvement of the technical solution, the data analysis module includes an abnormality analysis unit, and the process of the abnormality analysis unit determining abnormal features in employee attendance records specifically includes:

[0030] The abnormal scores of all employee attendance record data points are counted, and the abnormal threshold is determined using the quantile method. The abnormal threshold is used to distinguish normal points from abnormal points. The employee attendance record data points corresponding to the abnormal scores greater than the abnormal threshold are regarded as abnormal data points. And according to the shortest path length corresponding to the abnormal data point, the corresponding isolated tree in the isolation forest is found as the abnormal feature isolated tree;

[0031] According to the shortest path in the abnormal feature isolation tree, the features and split points on the path are recorded, and the frequency of each feature in the shortest path is recorded. The features are sorted from high to low according to the frequency of occurrence, and the feature with the highest frequency of occurrence is selected as the abnormal feature.

[0032] As a further improvement of the present technical solution, the attendance management module summarizes the employee information, abnormal score and abnormal features of each abnormal data point into an abnormal attendance report; generates targeted management suggestions based on the abnormal features, and generates corresponding management suggestions through the rule engine in the public database based on the abnormal features and the frequency of occurrence of the abnormal features.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] This enterprise attendance management system based on big data analysis combines the shortest path length and average path length of data points in all isolated trees to form a comprehensive path length and calculate the anomaly score, thereby capturing abnormal features more comprehensively; after determining the abnormal data points, the system further analyzes the shortest path, extracts and counts the features and split points on the path, and determines the specific features that cause the anomaly; finally, the rule engine generates management suggestions for specific abnormal features to help management take more scientific and effective management measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a schematic diagram of the overall module of the present invention;

[0036] Figure 2 It is a schematic diagram of the data analysis module unit of the present invention.

[0037] In the figure: 100, data collection module; 200, data analysis module; 201, model building unit; 202, path analysis unit; 203, abnormal score calculation unit; 204, abnormal analysis unit; 300, attendance management module. DETAILED DESCRIPTION

[0038] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0039] Next, see Figure 1-Figure 2 The present invention provides a technical solution: an enterprise attendance management system based on big data analysis, including a data collection module 100, a data analysis module 200 and an attendance management module 300.

[0040] The data collection module 100 collects employee attendance records from different data sources, where the different data sources include but are not limited to time clock machines, face recognition devices, fingerprint recognition devices and internal enterprise systems (such as HR systems, ERP systems); employee attendance records include but are not limited to multiple attendance features such as time characteristics, location characteristics, abnormal behavior characteristics and working day characteristics.

[0041] The model building unit 201 in the data analysis module 200 is used to analyze the employee attendance records and build an isolation forest algorithm model based on the employee attendance records, wherein the isolation forest algorithm model includes multiple isolated trees, specifically including:

[0042] Randomly select an attendance feature from different attendance features in the employee attendance record, and randomly select a split point within the value range of the selected attendance feature;

[0043] The data set is divided into two subsets according to the split point, one subset contains data points with eigenvalues ​​less than the split point, and the other subset contains data points with eigenvalues ​​greater than or equal to the split point;

[0044] Recursively perform the above steps for each subset until the number of data points in the subset is less than the preset threshold or the maximum tree depth is reached;

[0045] According to the above steps, multiple isolated trees are repeatedly constructed to form an isolation forest.

[0046] The path analysis unit 202 in the data analysis module 200 calculates the shortest path length and average path length of each employee data in all isolated trees, specifically including:

[0047] For each isolated tree in the isolation forest, input each employee attendance record data point;

[0048] Starting from the root node, recursively assign data points to corresponding child nodes according to features and split points until the data points reach the leaf node;

[0049] Record the path length of the data point from the root node to the leaf node;

[0050] Store the path length of each employee attendance record data point in all isolated trees in a matrix, where the rows of the matrix correspond to the data points and the columns correspond to the isolated trees;

[0051] For each employee attendance record data point, find the minimum value from the corresponding row of the path length matrix, that is, the shortest path of the data point in all isolated trees, as the shortest path length;

[0052] For each employee attendance record data point, the average of all path lengths is calculated from the corresponding row of the path length matrix by summing the extracted path lengths and then dividing by the number of isolated trees as the average path length.

[0053] By calculating the shortest path length and average path length of each employee attendance record data point in all isolated trees, the abnormality of the data point can be evaluated more comprehensively. The shortest path length reflects the isolation degree of the data point in a certain tree, while the average path length reflects the average isolation degree of the data point in all trees. Combining the shortest path length and the average path length can better explain the abnormal characteristics of the data point and help business personnel understand and deal with these abnormal points. Due to the use of multiple isolated trees, the calculation results of the shortest path length and the average path length are more stable, reducing the impact of the error of a single tree on the final detection result.

[0054] The abnormality score calculation unit 203 in the data analysis module 200 calculates the abnormality score of each employee attendance record data point according to the shortest path length and the average path length, specifically including:

[0055] The shortest path length and the average path length are combined to form a comprehensive path length, which is calculated using the weighted sum method. The formula is as follows:

[0056] ,in represents the comprehensive path length, represents the shortest path length; represents the average path length; and Respectively represent the weights of the shortest path length and the average path length, which are used to adjust according to actual needs;

[0057] The formula for calculating the abnormal score of each employee's attendance record data point is as follows:

[0058] ,in represents the abnormality score, It is a constant representing the expected path length of an isolated tree, which is ;

[0059] in, is the number of data points.

[0060] The abnormality analysis unit 204 in the data analysis module 200 determines whether the employee attendance is abnormal according to the abnormality score, and determines the abnormal features in the employee attendance record according to the shortest path length, specifically including:

[0061] The abnormal scores of all employee attendance record data points are counted, and the abnormal threshold is determined using the quantile method. For example, the 95% quantile is used as the abnormal threshold, which is used to distinguish normal points from abnormal points. The employee attendance record data points corresponding to the abnormal scores greater than the abnormal threshold are used as abnormal data points. And according to the shortest path length corresponding to the abnormal data point, the corresponding isolated tree in the isolation forest is found as the abnormal feature isolated tree;

[0062] According to the shortest path in the abnormal feature isolation tree, the features and split points on the path are recorded, and the frequency of each feature in the shortest path is recorded. The features are sorted from high to low according to the frequency of occurrence, and the feature with the highest frequency of occurrence is selected as the abnormal feature.

[0063] The attendance management module 300 is used to generate abnormal attendance reports based on abnormal characteristics in employee attendance records and provide management suggestions, including:

[0064] Summarize the employee information, anomaly score and anomaly features of each abnormal data point into an abnormal attendance report;

[0065] Generate targeted management suggestions based on abnormal features. Generate corresponding management suggestions based on abnormal features and the frequency of abnormal features through the rule engine in the public database, for example:

[0066] When the abnormal feature is absence (more than 3 days of absence) and the frequency is high (for example, the number of occurrences exceeds a certain threshold, such as 15): it is recommended that the human resources department pay attention to the health of employees and understand whether medical support or psychological counseling is needed; for employees who are absent for a long time, consider adjusting their work arrangements or positions;

[0067] Medium frequency (e.g., occurrences between 10-15 times): It is recommended that department managers interview employees to understand the reasons for absence and record them; provide necessary support for reasonable absences (e.g., sick leave) and issue warnings for unreasonable absences (e.g., absences without reason);

[0068] Low frequency (e.g., occurrences between 5 and 10): It is recommended that department managers pay attention to employees’ work pressure and provide appropriate stress relief measures; for individual absences, flexible handling should be provided, such as providing opportunities for compensatory leave.

[0069] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An enterprise attendance management system based on big data analysis, characterized in that: It comprises a data collection module (100), a data analysis module (200) and an attendance management module (300), wherein: The data collection module (100) collects employee attendance records; The data analysis module (200) is used to analyze employee attendance records, and to construct an isolation forest algorithm model based on the employee attendance records, wherein the isolation forest algorithm model includes multiple isolated trees; to calculate the shortest path length and average path length of each employee data in all isolated trees; to calculate the abnormality score of each data point based on the shortest path length and the average path length; to determine whether the employee attendance is abnormal based on the size of the abnormality score, and to determine the abnormal features in the employee attendance record based on the shortest path length; The attendance management module (300) is used to generate an abnormal attendance report based on abnormal features in the employee attendance records, and provide management suggestions.

2. The enterprise attendance management system based on big data analysis according to claim 1 is characterized in that: The data collection module (100) collects employee attendance records from different data sources, wherein the employee attendance records include multiple attendance features.

3. The enterprise attendance management system based on big data analysis according to claim 2 is characterized in that: The data analysis module (200) comprises a model building unit (201), and the process of the model building unit (201) building an isolation forest algorithm model specifically comprises: Randomly select an attendance feature from different attendance features in the employee attendance record, and randomly select a split point within the value range of the selected attendance feature; The data set is divided into two subsets according to the split point, one subset contains data points with eigenvalues ​​less than the split point, and the other subset contains data points with eigenvalues ​​greater than or equal to the split point; Recursively perform the above steps for each subset until the number of data points in the subset is less than the preset threshold or the maximum tree depth is reached; According to the above steps, multiple isolated trees are repeatedly constructed to form an isolation forest.

4. The enterprise attendance management system based on big data analysis according to claim 3 is characterized in that: The data analysis module (200) comprises a path analysis unit (202), and the process of calculating the path length by the path analysis unit (202) specifically comprises: For each isolated tree in the isolation forest, input each employee attendance record data point; Starting from the root node, recursively assign data points to corresponding child nodes according to features and split points until the data points reach the leaf node; Record the path length of the data point from the root node to the leaf node; The path length of each employee attendance record data point in all isolated trees is stored in a matrix to obtain a path length matrix, where the rows of the matrix correspond to the data points and the columns correspond to the isolated trees.

5. The enterprise attendance management system based on big data analysis according to claim 4 is characterized in that: The path analysis unit (202) calculates the shortest path length by finding the minimum value from the corresponding row of the path length matrix to determine the shortest path of each employee attendance record data point in all isolated trees.

6. The enterprise attendance management system based on big data analysis according to claim 5 is characterized in that: The path analysis unit (202) calculates the average path length by calculating the average of all path lengths from the corresponding row of the path length matrix by summing the extracted path lengths and then dividing by the number of isolated trees.

7. The enterprise attendance management system based on big data analysis according to claim 6 is characterized in that: The data analysis module (200) comprises an abnormality score calculation unit (203), and the abnormality score calculation process of the abnormality score calculation unit (203) specifically comprises: The shortest path length and the average path length are combined to form a comprehensive path length, which is calculated using the weighted sum method. The formula is as follows: ,in represents the comprehensive path length, represents the shortest path length; represents the average path length; and Represent the weights of the shortest path length and the average path length respectively; The formula for calculating the abnormal score of each employee's attendance record data point is as follows: ,in represents the abnormality score, It is a constant representing the expected path length of an isolated tree, which is ; in, is the number of data points.

8. The enterprise attendance management system based on big data analysis according to claim 7 is characterized in that: The data analysis module (200) comprises an abnormality analysis unit (204), and the process of the abnormality analysis unit (204) determining abnormal features in employee attendance records specifically comprises: The abnormal scores of all employee attendance record data points are counted, and the abnormal threshold is determined using the quantile method. The abnormal threshold is used to distinguish normal points from abnormal points. The employee attendance record data points corresponding to the abnormal scores greater than the abnormal threshold are regarded as abnormal data points. And according to the shortest path length corresponding to the abnormal data point, the corresponding isolated tree in the isolation forest is found as the abnormal feature isolated tree; According to the shortest path in the abnormal feature isolation tree, the features and split points on the path are recorded, and the frequency of each feature in the shortest path is recorded. The features are sorted from high to low according to the frequency of occurrence, and the feature with the highest frequency of occurrence is selected as the abnormal feature.

9. The enterprise attendance management system based on big data analysis according to claim 8 is characterized in that: The attendance management module (300) summarizes the employee information, abnormal score and abnormal features of each abnormal data point into an abnormal attendance report; generates targeted management suggestions based on the abnormal features, and generates corresponding management suggestions based on the abnormal features and the frequency of occurrence of the abnormal features through a rule engine in a public database.

Images