Segmented score analysis method based on ranking and average score

Abstract

The invention discloses a staged score analysis method based on ranking and average score, which comprises the following steps of: by constructing a student learning portrait based on historical ranking sequence variance and linear regression slope, extending an analysis perspective from single static data to a long-term dynamic trend. The system can automatically identify different types of students such as a stable type, a progressive type, a backward type, a fluctuation type and the like, and generates personalized comments and analysis reports fitting own development characteristics for the students. Therefore, students and parents are helped to clearly understand the real state of the learning track, a key basis for accurately identifying potential problems and implementing differentiated tutoring can be provided for teachers, and excitation, diagnosis and guiding functions of evaluation are fully exerted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of big data analysis, and particularly relates to a segmented score analysis method based on ranking and average score. BACKGROUND

[0002] Traditional score analysis methods usually rely on fixed score segmentation mechanisms, such as the commonly used 60-point passing line and 90-point excellent line standard, or simple ordering comparison based on test ranking. Although these methods are simple to implement, they have significant technical limitations in actual teaching evaluation.

[0003] Firstly, the fixed segmentation mode cannot adapt to the inherent characteristics of different disciplines. There are essential differences in knowledge structure, ability assessment focus and scoring standards between liberal arts and science disciplines, leading to systematic differences in their score distribution characteristics. For example, the scores of subjects such as mathematics and physics usually have a large standard deviation and discrimination, and the high-segment students' score distribution is dispersed; while the score distribution of subjects such as Chinese and history is relatively concentrated, and the density of middle-segment students is higher. Using a unified fixed segmentation standard cannot accurately reflect this subject difference, leading to distorted analysis results and making it difficult to truly reflect the teaching quality and learning effect of each subject.

[0004] Secondly, the existing technology pays too much attention to static and single test results, and lacks dynamic tracking and evaluation of students' learning development trends. Simple score or ranking comparison cannot identify students' long-term learning behavior patterns. For teaching managers, it is difficult to intervene in teaching and provide individualized guidance based on this situation; for students, they cannot obtain scientific evaluation that matches their effort level and development trajectory, which weakens the incentive and guidance role of the evaluation system. SUMMARY

[0005] The present application overcomes the shortcomings of the prior art and provides a segmented score analysis method based on ranking and average score.

[0006] To achieve the above purpose, the technical solution adopted by the present application is as follows: a segmented score analysis method based on ranking and average score, comprising the following steps:

[0007] S1, obtaining the score data of each student in the current test from the score database, and obtaining the ranking data of each student in the previous tests within a specified time range from the historical archives;

[0008] S2, for each subject of the current examination, calling a subject analysis module, retrieving a historical performance data set of the subject from the historical archive, calculating a standard deviation of the historical performance data set as a dispersion index, and calculating a difference between a high-score group average and a low-score group average in the historical performance data set as a discrimination index; inputting the dispersion index and the discrimination index into a predefined segmentation weight generation algorithm, and outputting a set of dynamic segmentation weight coefficients for defining subject performance intervals;

[0009] S3, for each student, calling a profile construction module, extracting ranking data of the past N examinations to form a personal historical ranking sequence, calculating a variance of the personal historical ranking sequence as a stability index, and calculating a linear regression slope of the last M rankings as a change trend index; inputting the stability index and the change trend index into a predefined classification rule set for matching, and outputting a learning profile type of the student;

[0010] S4, segmenting and mapping subject performance of all students using the set of dynamic segmentation weight coefficients obtained in step S2 to obtain subject static segmentation position data of each student; calling the learning profile type obtained in step S3, selecting a corresponding evaluation strategy according to different types to perform semantic processing and context modification on the subject static segmentation position data, and generating personalized comment data; combining and packaging the subject static segmentation position data, the learning profile type, and the personalized comment data to generate a personalized analysis report.

[0011] In a preferred embodiment of the present application, in step S2, the process of calculating the difference between the high-score group average and the low-score group average in the historical performance data set comprises:

[0012] S201, sorting all performance samples in the historical performance data set from high to low according to scores;

[0013] S202, selecting performance samples ranked in the top first preset percentage to form a high-score group, and calculating an average thereof as a high-score group average;

[0014] S203, selecting performance samples ranked after the first preset percentage to form a low-score group, and calculating an average thereof as a low-score group average;

[0015] S204, the difference between the high-score group average and the low-score group average as the discrimination index.

[0016] In a preferred embodiment of the present application, in step S2, the calculation process of the predefined segmentation weight generation algorithm comprises:

[0017] S211, set a segmentation strategy template for different types of subjects in advance, the segmentation strategy template including a science subject template and an art subject template;

[0018] S212, after the segmentation weight generation algorithm receives the dispersion index and the discrimination index, first, according to the numerical value size, determine whether the target subject tends to be more scientific or more artistic, and select the corresponding segmentation strategy template;

[0019] S213, based on the selected segmentation strategy template as a framework, use the numerical value of the dispersion index to scale the interval span of each partition by equal ratio, and use the numerical value of the discrimination index to fine-tune the threshold value of each partition interval, to generate a dynamic segmentation weight coefficient set including multiple score intervals and corresponding weight coefficients.

[0020] In a preferred embodiment of the present application, in step S3, the matching process of the pre-defined classification rule set is:

[0021] S301, if the stability index is lower than a first preset threshold value and the change trend index has an absolute value lower than a second preset threshold value, it is determined that the learning portrait type is stable;

[0022] S302, if the stability index is lower than the first preset threshold value and the change trend index is a negative value and has an absolute value higher than the second preset threshold value, it is determined that the learning portrait type is progressive;

[0023] S303, if the stability index is lower than the first preset threshold value and the change trend index is a positive value and has an absolute value higher than the second preset threshold value, it is determined that the learning portrait type is regressive;

[0024] S304, if the stability index is higher than the first preset threshold value, regardless of the value of the change trend index, it is determined that the learning portrait type is fluctuant.

[0025] In a preferred embodiment of the present application, in step S4, the selection of the corresponding evaluation strategy according to different types includes:

[0026] S401, when the learning portrait type is progressive, the selected evaluation strategy includes: comparing the subject static segmentation position data with the historical segmentation position data of the student in the last test, if the current segmentation level is higher, adding a positive semantic label for identifying the progress amplitude when generating the personalized comment data;

[0027] S402, when the learning portrait type is stable, the selected evaluation strategy includes: comparing the subject static segmentation position data with the overall segmentation distribution data of the group to which the student belongs, and generating the personalized comment data for describing the current position stability and relative level;

[0028] S403, when the learning portrait type is regression or fluctuation, the selected evaluation strategy includes: triggering a warning logic, generating a suggestive sentence containing a suggested knowledge point in the personalized comment data, and preferentially pushing the personalized analysis report containing a warning identifier to a teacher management terminal.

[0029] In a preferred embodiment of the present application, the segmented achievement analysis method further comprises:

[0030] The learning portrait type data of all students is received, statistical calculation is performed, and distribution ratio data of each portrait type in a class or grade range is obtained; at the same time, the subject static segmentation position data of all students is received, and the number distribution data of each score segment is generated by aggregation according to the interval defined by the dynamic segmentation weight coefficient set; the distribution ratio data and the number distribution data are integrated to generate a group analysis report for teachers.

[0031] In a preferred embodiment of the present application, the value range of the front first preset percentage and the rear second preset percentage is 15% to 20%.

[0032] In a preferred embodiment of the present application, the science subject template is configured to assign a higher weight subdivision density to the high score interval, and the arts subject template is configured to assign a higher weight subdivision density to the middle score interval.

[0033] In a preferred embodiment of the present application, the achievement data obtained from the achievement database server is cleaned to process missing values or abnormal values; the ranking data obtained from the historical archive server is subjected to consistency verification to ensure that the ranking calculation yardstick of all previous examinations is uniform; and the cleaned and verified data is stored in a temporary cache database for subsequent step calling.

[0034] In a preferred embodiment of the present application, in the step S4, the process of generating a personalized analysis report includes: calling a report template library, which stores text segment templates matched with different learning portrait types and different subject static segmentation position data; selecting corresponding text segments from the report template according to the specific data of the current student for combination; inserting the personalized comment data into a predetermined position in the combined text segments, and finally rendering a complete personalized analysis report document containing data charts and text analysis.

[0035] The generation and maintenance method of the text segment template in the report template library comprises: collecting typical comments written by education experts for different student types and achievement levels to establish an original corpus; analyzing the original corpus by using natural language processing technology to extract key evaluation dimensions and high-frequency words; based on the analysis results, a set of structured text segment templates containing replaceable variable placeholders are constructed; the text segment templates are updated and optimized through the administrator terminal according to new education policies or evaluation ideas.

[0036] The present application solves the defects in the background art, and has the following advantages:

[0037] (1) By introducing the standard deviation of subject history score data and the average score difference of high and low groups as quantitative basis, and dynamically generating segmented weight coefficients in accordance with the specific subject score distribution law based on these indicators, the disadvantages of the traditional fixed segmentation mode are overcome. The science high segmentation is more fine, the arts segmentation analysis is more focused, and the analysis results can truly and accurately reflect the score distribution characteristics of different subjects, providing a scientific and reliable data basis for teaching evaluation.

[0038] (2) By constructing a student learning portrait based on stability index and change trend index, the analysis perspective is extended from single static data to long-term dynamic trend. The system can automatically identify different types of students and generate personalized comments and analysis reports that fit their own development characteristics. Further, it not only helps students and parents clearly understand the true state of the learning trajectory, but also provides teachers with key evidence for accurately identifying potential problems and implementing differentiated coaching, fully realizing the evaluation functions of motivation, diagnosis and guidance. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, a brief introduction will be given below to the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments described in the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor;

[0040] Figure 1 is the implementation flowchart of the segmented score analysis method of the preferred embodiment of the present application. DETAILED DESCRIPTION

[0041] With reference to the drawings and embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0042] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application. However, the present application can be practiced without the specific details, other than those described herein, and therefore the scope of the present application is not limited to the details of the embodiments described herein.

[0043] As shown in the figure, the segmented performance analysis method based on ranking and average score includes the following steps: Figure 1

[0044] S1, obtaining the performance data of each student in the current examination from the performance database, and obtaining the ranking data of each student in the previous examinations within a specified time range from the historical archives.

[0045] S2, for each subject of the current examination, calling the subject analysis module, retrieving the historical performance data set of the subject from the historical archives, calculating the standard deviation of the historical performance data set as a dispersion index, and calculating the difference between the average score of the high-score group and the average score of the low-score group in the historical performance data set as a discrimination index; inputting the dispersion index and the discrimination index into the pre-defined segmented weight generation algorithm, and outputting a set of dynamic segmented weight coefficients for defining the performance intervals of the subject.

[0046] Further, in step S2, the process of calculating the difference between the average score of the high-score group and the average score of the low-score group in the historical performance data set includes:

[0047] S201, sorting all performance samples in the historical performance data set from high to low according to the scores;

[0048] S202, selecting the top first preset percentage of performance samples to form a high-score group, and calculating the average score thereof as the average score of the high-score group;

[0049] S203, selecting the bottom first preset percentage of performance samples to form a low-score group, and calculating the average score thereof as the average score of the low-score group;

[0050] S204, the difference between the average score of the high-score group and the average score of the low-score group as the discrimination index.

[0051] The value range of the top first preset percentage and the bottom second preset percentage is 15% to 20%.

[0052] ​Further, in step S2, the calculation process of the pre-defined segmentation weight generation algorithm includes:

[0053] S211, pre-set segmentation strategy templates for different types of subjects, including science subject templates and arts subject templates; wherein the science subject templates are configured to assign higher weight subdivision density to high score intervals, and the arts subject templates are configured to assign higher weight subdivision density to middle score intervals.

[0054] S212, after receiving the dispersion index and the discrimination index, the segmentation weight generation algorithm first determines whether the target subject tends to be more scientific or more artistic according to the numerical value, and selects the corresponding segmentation strategy template;

[0055] S213, based on the selected segmentation strategy template as the framework, the numerical value of the dispersion index is used to scale the interval span of each partition, and the numerical value of the discrimination index is used to fine-tune the threshold value of each partition interval, to generate a dynamic segmentation weight coefficient set containing multiple score intervals and their corresponding weight coefficients.

[0056] S3, for each student, call the portrait construction module, extract the ranking data of the past N times of examination to form a personal historical ranking sequence, calculate the variance of the personal historical ranking sequence as the stability index, and calculate the linear regression slope of the last M times of ranking as the change trend index; input the stability index and the change trend index into the pre-defined classification rule set for matching, and output the learning portrait type of the student.

[0057] Further, in step S3, the matching process of the pre-defined classification rule set is:

[0058] S301, if the stability index is lower than the first preset threshold and the absolute value of the change trend index is lower than the second preset threshold, the learning portrait type is determined to be stable;

[0059] S302, if the stability index is lower than the first preset threshold, the change trend index is negative, and the absolute value is higher than the second preset threshold, the learning portrait type is determined to be progressive;

[0060] S303, if the stability index is lower than the first preset threshold, the change trend index is positive, and the absolute value is higher than the second preset threshold, the learning portrait type is determined to be regressive;

[0061] S304, if the stability index is higher than the first preset threshold, regardless of the value of the change trend index, the learning portrait type is determined to be fluctuant.

[0062] S4, segmenting and mapping the subject scores of all students using the dynamic segmented weight coefficient set obtained in step S2 to obtain subject static segmented position data of each student; calling the learning portrait type obtained in step S3, selecting a corresponding evaluation strategy according to different types to perform semantic processing and context modification on the subject static segmented position data, and generating personalized comment data; combining and packaging the subject static segmented position data, the learning portrait type and the personalized comment data to generate a personalized analysis report.

[0063] Further, in step S4, the corresponding evaluation strategy is selected according to different types, which includes:

[0064] S401, when the learning portrait type is the progress type, the selected evaluation strategy includes: comparing the subject static segmented position data with the historical segmented position data of the student in the last test, if the current segmented level is higher, adding a positive semantic label for identifying the progress amplitude when generating the personalized comment data;

[0065] S402, when the learning portrait type is the stable type, the selected evaluation strategy includes: comparing the subject static segmented position data with the overall segmented distribution data of the group to which the student belongs, and generating personalized comment data for describing the stability of the current position and the relative level;

[0066] S403, when the learning portrait type is the regression type or the fluctuation type, the selected evaluation strategy includes: triggering a warning logic, generating a suggestive sentence containing a suggestion for paying attention to a knowledge point in the personalized comment data, and preferentially pushing the personalized analysis report containing a warning identifier to a teacher management terminal.

[0067] In another preferred embodiment, the segmented score analysis method further includes:

[0068] Receiving learning portrait type data of all students, performing statistical calculation to obtain distribution proportion data of each portrait type in a class or grade range; at the same time, receiving subject static segmented position data of all students, aggregating according to intervals defined by the dynamic segmented weight coefficient set to generate person number distribution data of each score segment; integrating the distribution proportion data and the person number distribution data to generate a group analysis report for teachers.

[0069] Embodiment one: analysis of science scores

[0070] This embodiment takes the mathematics subject scores of a high school grade three mock exam as an example for illustration. The total score of the mathematics subject is 150, and the number of students participating in the exam is 200. The system obtains the original scores of all students in the mathematics subject in this exam from the score database, and retrieves the school-wide ranking data of the previous four large-scale exams in the current school year from the historical archives.

[0071] The subject analysis module calls the historical performance data set of the mathematics subject, which contains the mathematics performance of 12 examinations in the past three years, with a total sample size of 2400. The average scores of the 12 examinations are: 102.3, 105.7, 98.6, 107.2, 103.8, 100.5, 109.1, 104.6, 101.9, 106.4, 103.1, 108.2 (unit: points). The arithmetic mean of the 2400 performance data is

[0072] .

[0073] The standard deviation is 28.5 according to the 2400 original scores, indicating that the degree of performance dispersion is high.

[0074] Subsequently, the system sorts the historical performance, takes the top 20% of the performance sample to calculate the high score group average, and takes the last 20% of the performance sample to calculate the low score group average, which is:

[0075] Take the 200 performance of the last examination (the 12th) as the calculation sample, and sort the performance from high to low:

[0076] The top 20% (40 people) performance: 148, 147, 146, 145, 144, 143, 142, 141, 140, 139, 138, 137, 136, 135, 134, 133, 132, 131, 130, 129, 128, 127, 126, 125, 124, 123, 122, 121, 120, 119, 118, 117, 116, 115, 114, 113, 112, 111, 110, 109.

[0077] The high score group average is (148+147+...+109) / 40=5324 / 40=133.1 points.

[0078] The last 20% (40 people) performance: 69, 68, 67, 66, 65, 64, 63, 62, 61, 60, 59, 58, 57, 56, 55, 54, 53, 52, 51, 50, 49, 48, 47, 46, 45, 44, 43, 42, 41, 40, 39, 38, 37, 36, 35, 34, 33, 32, 31, 30.

[0079] The low score group average is (69+68+...+30) / 40=1980 / 40=49.5 points.

[0080] The discrimination index is 133.1-49.5=83.6 points.

[0081] The difference between the two was 83.6 points, which is a significant indicator of discrimination.

[0082] The segmented weight generation algorithm determines that mathematics exhibits typical science characteristics based on these two indicators and selects a science-related subject template. This template pre-divides the high-score range of 120-150 points into four intervals: 120-127, 128-135, 136-143, and 144-150. The system uses a standard deviation of 28.5 to proportionally scale the span of each interval and adjusts the interval thresholds based on a discrimination index of 45.2, ultimately generating a dynamic set of segmented weight coefficients.

[0083] Taking student A as an example, their math score this time was 129. The profile building module extracted the student's grade ranking sequence from the past five exams as [85, 78, 72, 65, 58].

[0084] The ranking sequence of the five exams is: [85, 78, 72, 65, 58].

[0085] average value .

[0086] Variance calculation: .

[0087] Variance = 449.2 / 4 = 112.3.

[0088] Linear regression slope calculation:

[0089] Let x be the number of times the exam was taken (1, 2, 3, 4, 5), and y be the ranking.

[0090] Slope: .

[0091] The linear regression slope for the three most recent rankings is calculated as follows: The absolute value of the score exceeded the preset threshold of 5. Based on the classification rules, the system determined the student to be an improved student. The comprehensive evaluation module mapped the score of 129 to the 128-135 range and retrieved the student's historical score range from the previous exam (located in the 112-119 range). Due to the improvement in the current score range, the system added a positive semantic tag indicating significant improvement when generating personalized comments. The final analysis report indicated that the student's math score of 129 placed them in the excellent 128-135 range, representing an improvement of one score range compared to the previous exam, demonstrating a continuous improvement trend.

[0092] Example 2: Analysis of Humanities Scores

[0093] This example is illustrated by the same test Chinese subject performance. Chinese subject total score of 150 points, reference students with 200. The system synchronously access Chinese raw score and historical ranking data.

[0094] Subject analysis module processing Chinese subject in the last three years 12 times of the historical performance data, the total sample size of 2400. 12 test average scores were: 108.5, 107.8, 109.2, 106.9, 108.1, 107.5, 109.8, 108.3, 107.2, 108.9, 107.6, 109.1 (unit: points).

[0095] Standard deviation calculation process:

[0096] The arithmetic mean of the total 2400 performance data .

[0097] Variance: .

[0098] If the 2400 original score is calculated, the standard deviation is 12.3, and the dispersion degree is relatively low. Take the top 15% of the performance sample to calculate the high score group average, and the last 15% of the performance sample to calculate the low score group average, which is:

[0099] Take the last test (12th) of 200 Chinese performance as a sample:

[0100] The top 15% (30 people) performance: 142, 141, 140, 139, 138, 137, 136, 135, 134, 133, 132, 131, 130, 129, 128, 127, 126, 125, 124, 123, 122, 121, 120, 119, 118, 117, 116, 115, 114, 113.

[0101] High score group average=(142+141+...+113) / 30=3825 / 30=127.5 points.

[0102] The last 15% (30 people) performance: 95, 94, 93, 92, 91, 90, 89, 88, 87, 86, 85, 84, 83, 82, 81, 80, 79, 78, 77, 76, 75, 74, 73, 72, 71, 70, 69, 68, 67, 66.

[0103] Low score group average=(95+94+...+66) / 30=2415 / 30=80.5 points.

[0104] Discrimination index=127.5-80.5=47.0 points.

[0105] The algorithm determines that the subject has liberal arts characteristics and selects a liberal arts subject template. The template is configured to refine the score segment 90-120, and is preset to have five intervals: 90-96, 97-103, 104-110, 111-117, and 118-125. The system adjusts the interval span using a standard deviation of 12.3, fine-tunes the threshold value based on a discrimination degree of 18.5, and generates a dynamic segmentation weight coefficient set suitable for Chinese subjects.

[0106] Taking student B as an example, the Chinese score is 108. The portrait construction module obtains the past five test ranking sequences of the student as [32, 35, 33, 34, 36].

[0107] Five test ranking sequences: [32, 35, 33, 34, 36].

[0108] Mean .

[0109] Variance calculation: .

[0110] Variance = 10 / 4 = 2.5.

[0111] Linear regression slope calculation:

[0112] Let x be the number of tests (1, 2, 3, 4, 5) and y be the ranking.

[0113] Slope .

[0114] The calculated variance is 2.5, which is much lower than the preset threshold of 150; the linear regression slope is 0.7, and the absolute value is lower than the preset threshold of 5. The system determines that the student is stable. The comprehensive evaluation module maps 108 to the 104-110 segment, and after comparing the position data with the overall distribution of the class, it finds that it is in the upper middle level and extremely stable. The generated personalized comment is: Chinese score 108, stable in the 104-110 good interval, the score remains in a stable state, and the class ranking always maintains within the top 20%. The system does not trigger the warning mechanism, and generates a regular analysis report and pushes it to the student end.

[0115] At the same time, the system aggregates and analyzes the data of all 200 students to generate a group analysis report for teachers: the number of students in the high score segment (136-150) of mathematics accounts for 18.5%, and the number of students in the high score segment (118-150) of Chinese accounts for 22.3%; the proportion of progress type students in mathematics is 24.5%, and in Chinese is 18.7%, fully reflecting the characteristics of science subjects that can better identify progress trends after dynamic segmentation.

[0116] The above is based on the ideal embodiment of the application, through the above description, relevant personnel can make various changes and modifications without deviating from the technical idea of the application. The technical scope of the application is not limited to the content of the specification, and the technical scope must be determined according to the scope of claims.

Claims

1. A method for analyzing segmental performance based on ranking and average score, characterized in that, The method comprises the following steps: S1, obtaining student score data of a current examination from a score database, and obtaining ranking data of each student in previous examinations within a specified time range from a historical archive; S2, for each subject of the current examination, calling a subject analysis module, retrieving a historical score data set of the subject from the historical archive, calculating the standard deviation of the historical score data set as a dispersion index, and calculating the difference between the average score of the high-score group and the average score of the low-score group in the historical score data set as a discrimination index; inputting the dispersion index and the discrimination index into a pre-defined segmentation weight generation algorithm, and outputting a set of dynamic segmentation weight coefficients for defining subject score intervals; S3, for each student, calling a portrait construction module, extracting ranking data of the past N examinations to form a personal historical ranking sequence, calculating the variance of the personal historical ranking sequence as a stability index, and calculating the linear regression slope of the last M rankings as a change trend index; inputting the stability index and the change trend index into a pre-defined classification rule set for matching, and outputting the student's learning portrait type; S4, using the set of dynamic segmentation weight coefficients obtained in step S2 to segment and map the subject scores of all current students, to obtain subject static segmentation position data of each student; calling the learning portrait type obtained in step S3, selecting a corresponding evaluation strategy according to different types to perform semantic processing and context modification on the subject static segmentation position data, and generating personalized comment data; combining and packaging the subject static segmentation position data, the learning portrait type, and the personalized comment data to generate a personalized analysis report.

2. The ranking and average-based partial performance analysis method according to claim 1, wherein: In step S2, the process of calculating the difference between the average score of the high-score group and the average score of the low-score group in the historical score data set comprises: S201, sorting all score samples in the historical score data set from high to low according to scores; S202, selecting the top first preset percentage of score samples to form a high-score group, and calculating the average score as the average score of the high-score group; S203, selecting the last first preset percentage of score samples to form a low-score group, and calculating the average score as the average score of the low-score group; S204, the difference between the average score of the high-score group and the average score of the low-score group is taken as the discrimination index.

3. The ranking and average based segmental performance analysis method according to claim 1, wherein: In step S2, the calculation process of the pre-defined segmentation weight generation algorithm comprises: S211, pre-setting segmentation strategy templates for different types of subjects, the segmentation strategy templates including science subject templates and arts subject templates; S212, after the segmentation weight generation algorithm receives the dispersion index and the discrimination index, it first determines whether the target subject tends to be more similar to a science characteristic or an arts characteristic according to the numerical value, and selects the corresponding segmentation strategy template; S213. Based on the selected segmentation strategy template, the interval span of each partition is scaled proportionally using the value of the dispersion index, and the threshold of each partition interval is fine-tuned using the value of the discrimination index, thereby generating a dynamic segmentation weight coefficient set containing multiple score intervals and their corresponding weight coefficients.

4. The ranking and average based segmental performance analysis method according to claim 1, wherein: In step S3, the matching process of the predefined classification rule set is as follows: S301. If the stability index is lower than the first preset threshold and the absolute value of the change trend index is lower than the second preset threshold, then the learning profile type is determined to be stable. S302. If the stability index is lower than the first preset threshold and the trend index is negative and its absolute value is higher than the second preset threshold, then the learning profile type is determined to be progressive. S303. If the stability index is lower than the first preset threshold and the trend index is positive and its absolute value is higher than the second preset threshold, then the learning profile type is determined to be regressive. S304. If the stability index is higher than the first preset threshold, then regardless of the value of the change trend index, the learning profile type is determined to be fluctuating.

5. The ranking and average based segmental score analysis method according to claim 4, wherein: In step S4, selecting the corresponding evaluation strategy based on different types includes: S401. When the learning profile type is progressive, the selected evaluation strategy includes: comparing the static segment position data of the subject with the historical segment position data of the student's last exam; if the current segment level is higher, then adding a positive semantic label to identify the degree of progress when generating the personalized comment data. S402. When the learning profile type is stable, the selected evaluation strategy includes: comparing the static segmented position data of the subject with the overall segmented distribution data of the student's group to generate the personalized evaluation data used to describe the stability and relative level of its current position. S403. When the learning profile type is regressive or fluctuating, the selected evaluation strategy includes: triggering early warning logic, generating a prompt statement containing suggested knowledge points in the personalized comment data, and prioritizing the push of the personalized analysis report containing the early warning indicator to the teacher management terminal.

6. The ranking and average-based partial performance analysis method according to claim 1, wherein: The segmented score analysis method also includes: The system receives learning profile type data from all students, performs statistical calculations to obtain the distribution ratio data of each profile type within the class or grade; simultaneously, it receives the static segmentation position data of all students by subject, aggregates it according to the interval defined by the dynamic segmentation weight coefficient set, and generates the number distribution data of each score segment; the distribution ratio data and the number distribution data are integrated to generate a group analysis report for teachers.

7. The ranking and average-based partial performance analysis method according to claim 2, wherein: The values ​​of the first preset percentage and the second preset percentage are both within the range of 15% to 20%.

8. The ranking and average-based partial performance analysis method according to claim 3, wherein: The science subject templates are configured to assign higher weight subdivision density to the high score range, while the humanities subject templates are configured to assign higher weight subdivision density to the middle score range.

Images