KR102603784B1 - Modeling method for predicting risk of getting out of the way in online university - Google Patents

Abstract

A modeling method for predicting dropout risk groups through analysis of cyber university learning activity data is disclosed. As a means to solve the above-mentioned technical problem, the dropout risk group prediction modeling method through analysis of learning activity data of a cyber university according to an embodiment of the present invention collects primary variables and secondary variables respectively during a preset collection period. step, quantifying the collected primary and secondary variables, comparing the quantified primary variables with the standard level set for each variable to determine those scheduled to drop out during the semester, and quantifying the quantified secondary variables into each variable. It includes the step of determining those scheduled to drop out after the semester by comparing them with the standard level set for each semester.

Description

Modeling method for predicting dropout risk group through analysis of learning activity data at cyber university {Modeling method for predicting risk of getting out of the way in online university}

The present invention relates to a predictive modeling method for dropout risk groups through analysis of learning activity data at cyber universities. More specifically, the present invention relates to a learning method at cyber universities that minimizes student dropout by predicting in advance before students at online universities give up their studies. This is about a predictive modeling method for dropout risk groups through activity data analysis.

As information and communication technology develops, many lecture systems that used to be conducted offline are now conducted online. As the online lecture system develops, online lectures are not limited to simple lectures but are connected to the university system, and online universities, which allow students to participate in university lectures regardless of time and place, are gaining popularity.

In these online universities, most lectures are conducted online, so students must study independently. For this reason, online universities require a significant level of ICT (Information and Communications Technologies) capabilities compared to existing offline universities.

However, looking at the learning environment provided by current online universities, there is a significant lack of support from the surrounding environment for students who must study independently, and immediate feedback is not provided.

Additionally, due to the nature of online universities, students with very diverse academic backgrounds and levels enroll. For example, students who enroll in online universities include young students who have completed their regular education courses, but there are also many older students who were unable to complete their studies on time for one reason or another.

Likewise, in online universities, which show different characteristics from students at offline universities, the percentage of students who drop out in the first semester or immediately after admission reaches 20 to 26% of all freshmen.

Giving up on studies midway is a problem even for students who have carefully chosen to attend college, but it also leads to the problem of losing students to online universities. Therefore, in order to retain new students, online universities need to identify students who are unable to adapt to their studies or school before they leave and provide support measures.

Domestic Public Patent No. 10-20200051147 (published on May 13, 2020)

In order to solve the above-mentioned problems, the technical task to be achieved by the present invention is to track various learning activity data that affects learning for students at online universities, and to extract and identify students who are likely to give up their studies in advance. The purpose is to present a predictive modeling method for dropout risk groups through analysis of learning activity data at cyber universities to enable guidance.

The problems of the present invention are not limited to those mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below.

As a means to solve the above-mentioned technical problem, the dropout risk group prediction modeling method through analysis of learning activity data of a cyber university according to an embodiment of the present invention collects primary variables and secondary variables respectively during a preset collection period. step, quantifying the collected primary and secondary variables, comparing the quantified primary variables with the standard level set for each variable to determine those scheduled to drop out during the semester, and quantifying the quantified secondary variables into each variable. It includes the step of determining those scheduled to drop out after the semester by comparing them with the standard level set for each semester.

Preferably, the step of repeating the collecting and quantifying steps may be further included at each preset collection cycle.

Also, preferably, the step of determining those scheduled to drop out during the semester reflects and compares the results of repeated performance and determines that overlapping students are scheduled to drop out during the semester, and the step of determining those scheduled to drop out after the semester reflects the results of repeated performance. As a result of the comparison, overlapping students can be judged as those scheduled to drop out after the semester.

Also, preferably, the step of determining who is scheduled to drop out during the semester is to compare the primary variables step by step according to the preset variable order, extract students whose numerical values of the variables fall below each standard level at each step, and finally Students whose numerical values of variables are below the standard level can be judged to be scheduled to drop out during the semester.

Also, preferably, the step of determining those scheduled to drop out after the semester is to compare the secondary variables step by step according to the preset variable order, extract students whose numerical values of the variables fall below each standard level at each step, and finally Students whose numerical values of variables are below the standard level can be judged as scheduled to drop out after the semester.

Also preferably, the primary variable may include age, gender, entrance grade, quota classification, admission recruitment round, application fee, and basic tuition payment time.

Also preferably, the secondary variables may include age, gender, entrance grade, quota classification, admission recruitment round, application fee, basic tuition payment time, president's scholarship amount, national scholarship amount, course registration credits, and test application rate. .

According to the present invention, among the data appearing in the learning activities of an online university, variables related to the learning behavior of students who are willing to give up their studies are tracked, and by predicting those who are expected to give up their studies, a cyber system can accurately predict those who are expected to give up their studies. It is effective in providing a predictive modeling method for dropout risk groups through analysis of university learning activity data.

In addition, by accurately predicting those who are expected to give up their studies, efforts can be made to ensure that students continue their studies before they give up, which has the effect of minimizing student outflow.

The effects of the present invention are not limited to those mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description below.

1 is a block diagram of a prediction server according to a preferred embodiment of the present invention;
Figure 2 is a diagram illustrating the observation results of the age variable used in the prediction server according to a preferred embodiment of the present invention;
Figure 3 is a diagram illustrating the observation results of the gender variable used in the prediction server according to a preferred embodiment of the present invention.
Figure 4 is a diagram illustrating a method of predicting students scheduled to drop out during the semester through a step-by-step comparison procedure according to the order of variables according to a preferred embodiment of the present invention;
Figure 5 is a diagram illustrating a method of predicting those scheduled to drop out after a semester by a step-by-step comparison procedure according to the order of variables according to a preferred embodiment of the present invention;
Figure 6 is a flowchart illustrating a predictive modeling method for a dropout risk group through analysis of learning activity data at a cyber university according to a preferred embodiment of the present invention, and
Figure 7 is a flowchart illustrating a method for predicting a dropout risk group through analysis of learning activity data at a cyber university according to another preferred embodiment of the present invention.

The above objects, other objects, features and advantages of the present invention will be easily understood through the following preferred embodiments related to the attached drawings. However, the present invention is not limited to the embodiments described herein and may be embodied in other forms. Rather, the embodiments introduced herein are provided so that the disclosed content will be thorough and complete and so that the spirit of the present invention can be sufficiently conveyed to those skilled in the art.

In this specification, when an element is referred to as being on another element, it means that it may be formed directly on the other element or that a third element may be interposed between them. Additionally, in the drawings, the thickness of components is exaggerated for effective explanation of technical content.

In this specification, when terms such as first, second, etc. are used to describe components, these components should not be limited by these terms. These terms are merely used to distinguish one component from another. Embodiments described and illustrated herein also include complementary embodiments thereof.

Additionally, when a first element (or component) is referred to as being operated or executed on (ON) a second element (or component), the first element (or component) means that the second element (or component) is ON. It should be understood as being operated or executed in an environment in which it is operated or executed, or operated or executed through direct or indirect interaction with a second element (or component).

If any element, component, device, or system is said to contain a component consisting of a program or software, even if explicitly stated, that element, component, device, or system is not intended to allow that program or software to run or operate. It should be understood as including hardware (e.g., memory, CPU, etc.) or other programs or software (e.g., operating system or drivers required to run the hardware) required to run the computer.

In addition, unless specifically stated in the implementation of an element (or component), it should be understood that the element (or component) may be implemented in any form of software, hardware, or software and hardware.

Additionally, the terms used in this specification are for describing embodiments and are not intended to limit the present invention. As used herein, singular forms also include plural forms, unless specifically stated otherwise in the context. As used in the specification, 'comprises' and/or 'comprising' does not exclude the presence or addition of one or more other elements.

1 is a block diagram of a prediction server according to a preferred embodiment of the present invention.

The prediction server 100 is a server that predicts the risk of dropping out by analyzing data that appears through the online learning activities of students at an online university. Here, the dropout risk group refers to a group of students who give up their studies midway without completing the series of courses from admission to graduation from an online university. For example, dropping out includes giving up on studies during the semester and giving up on studies after the semester.

Referring to FIG. 1, the prediction server 100 according to a preferred embodiment of the present invention includes a network interface unit 110, a collection unit 120, a digitization unit 130, a prediction unit 140, and a storage unit 150. , and a control unit 160.

The network interface unit 110 supports a network interface between the prediction server 100 and external terminal devices (not shown). External terminal devices include, for example, terminal devices used by professors, staff, and students, such as Internet-enabled desktops, laptops, tablet PCs, and smartphones. Through the network interface unit 110, the prediction server 100 can transmit and receive various information with external terminal devices.

The collection unit 120 collects various information related to new and current students of the online university. Information collected in the collection unit 120 can be collected from student information basically provided by students of the online university, and can be collected from various bulletin boards on the online university's homepage through the network interface unit 110.

The collection unit 120 collects variables to be used in predicting students who are scheduled to give up their studies during a preset collection period. More specifically, the collection unit 120 may collect variables for a preset collection period once at a set collection time. Additionally, the collection unit 120 may repeatedly collect variables during the collection period several times at each preset collection cycle.

In this embodiment, the variables used to determine those scheduled to drop out during the semester are called primary variables, and the variables used to determine those scheduled to drop out after the semester are called secondary variables. Here, the primary variables may include age, gender, entrance grade, quota classification, admission recruitment round, screening fee, and basic tuition payment timing. In addition, secondary variables may include age, gender, entrance grade, quota classification, admission recruitment round, application fee, basic tuition payment time, president's scholarship amount, national scholarship amount, course registration credits, and test application rate.

In addition, the variables that can be collected by the collection unit 120 can be modified in more diverse ways. Variables related to academics may include attendance time, assignment progress rate, test taking rate, number of portal logins, academic notice confirmation, class participation, discussion participation, and class bulletin board utilization. Additionally, variables related to student background and characteristics may include age, gender, occupation, residential area, admission type, and selection qualifications. Additionally, variables related to academic conditions may include economic status, employment type, family situation, academic support, national scholarship, and school scholarship. In addition, variables related to social integration may include attendance rate at local meetings, attendance rate at offline events on campus, use of school SNS, and class bulletin board activity. Additionally, defining variables may include self-efficacy, school satisfaction, goal commitment, and stress index. Academic outcome variables may include school grades. These various variables can be applied to primary or secondary variables.

The numeric value unit 130 digitizes the primary variables and secondary variables collected by the collection part 120 and converts them into numerical values. To quantify the variables in the numeric value unit 130, a method appropriate for each variable must be used.

The age variable uses each age group or age ratio as a variable. The gender variable uses a nominal variable representing men and women. Occupational variables use nominal variables according to occupational group. The residential area variable uses a nominal variable according to the region. The admission type variable uses nominal variables based on new students, transfer students, etc. The selection qualification variables use nominal variables based on general selection, industry, military consignment, etc.

Attendance time variables are converted into course averages and progress rates. The assignment progress rate variable uses the submission rate for submitted assignments. The exam application rate variable uses the application rate for regular exams such as midterms and final exams. The portal login count variable uses the number of logins to the school portal. The academic notice confirmation variable is used by converting the confirmation of lecture notices for each subject into a ratio. The class participation variable uses the percentage of participation in the class bulletin board, discussion participation bulletin board, and question bulletin board.

National scholarship and on-campus scholarship variables use the amount of national scholarship and on-campus scholarship. The participation rate variable for offline events on campus uses the participation rate for various events such as entrance ceremonies, orientations, department outings, and sports competitions.

The prediction unit 140 extracts those scheduled to give up their studies using numerical values of various variables. Those scheduled to drop out of school extracted from the prediction unit 140 may include those scheduled to drop out during the semester and those scheduled to drop out after the semester.

In this embodiment, those scheduled to give up their studies (drop out mid-term) are divided into those scheduled to drop out during the semester and those scheduled to drop out after the semester, but this is not necessarily limited to this. In addition to the two examples already mentioned, those scheduled to give up their studies may be more subdivided or composed in various ways. For example, those scheduled to drop out after the semester can be further divided into those scheduled to drop out and those scheduled to take a leave of absence.

The prediction unit 140 can determine who is scheduled to drop out during the semester by comparing the numerical value of the primary variable with the standard level set for each variable, and compares the numerical value of the secondary variable with the standard level set for each variable. It is possible to determine who is scheduled to drop out after the semester based on their actions.

When the collection unit 120 repeatedly collects primary variables and secondary variables at each preset collection cycle, the prediction unit 140 sets each variable to the corresponding standard whenever the variables are repeatedly collected by the collection unit 120. It can be compared with the level, and students with overlapping comparison results can be judged as those scheduled to fail during the semester or those scheduled to fail after the semester.

In addition, the prediction unit 140 compares the primary variable or secondary variable step by step according to the preset variable order, extracts students whose numerical values of the variable are below each standard level at each step, and predicts the variable until the last step. Students whose numerical values are below the standard level can be judged as those scheduled to fail during the semester or those scheduled to fail after the semester.

The storage unit 150 stores all information necessary for the operation of the prediction server 100. For example, the storage unit 150 may store a preset reference level for each of the variables used in the operation of the prediction unit 140, and for the smooth operation of the prediction server 100, Basic information can be saved.

The control unit 160 controls the overall operation of the prediction server 100. That is, the control unit 160 controls signal input and output between the network interface unit 110, the collection unit 120, the digitization unit 130, the prediction unit 140, and the storage unit 150.

Figure 2 is a diagram illustrating the observation results of the age variable used in the prediction server according to a preferred embodiment of the present invention.

In this embodiment, among the various variables collected by the collection unit 120, the age variable is used to track and graph the re-enrollment rate for the next semester over three semesters. Students were divided by age and displayed on a graph into the group that dropped out during the semester (A), the group that dropped out after the semester (B), and the group that continued their studies (C).

(a) shows the change in re-enrollment rate by age group in the first semester when tracking began. Although the proportion of students who chose to drop out during the semester is not large, the proportion of students in their 20s appears to be the highest among various age groups.

(b) shows the change in re-enrollment rate by age group in the second semester after tracking began. It is not much different from the result in (a), but there was no dropout during the semester in the 60s age group.

(c) shows the change in re-enrollment rate by age group in the third semester after tracking began. Looking at the graph, the number of students who chose to drop out during the semester increased in all age groups, and in particular, the number of students who chose to drop out during the semester increased significantly among those in their 60s. Additionally, the number of students in their 20s who chose to drop out after the semester increased.

Referring to the graphs in (a) to (c), it is possible to identify the age group in which dropouts occur most during the semester or the age group in which dropouts occur most after the semester. By this, it is possible to manage the relevant age group to continue their studies.

Figure 3 is a diagram illustrating the observation results of the gender variable used in the prediction server according to a preferred embodiment of the present invention.

In this embodiment, among the various variables collected by the collection unit 120, the gender variable is used to track and graph the re-enrollment rate for the next semester over three semesters. As in the previous example, students were divided by age and displayed in a graph into a group that dropped out during the semester (A), a group that dropped out after the semester (B), and a group that continued their studies (C).

(a) shows the change in re-enrolment rate by gender in the first semester when tracking began. A small number of students, regardless of gender, appeared in the dropout group during the semester (A), and the dropout group after the semester (B) showed a small number of male students. This appeared a little more.

(b) shows the change in re-enrollment rate by gender in the second semester after tracking began. Compared to (a), slightly more students gave up their studies, and more male students both dropped out during the semester and dropped out after the semester. It appears that

(c) shows the change in re-enrolment rate by gender in the third semester after tracking began. Compared to (a) and (b), slightly more students gave up their studies, and still both dropped out during the semester and dropped out after the semester. appears to have more male students.

The graphs in (a) to (c) show that many male students choose to drop out after the semester, so schools need to take care to ensure that male students can continue their studies without giving up midway.

Figures 2 and 3 show the results of tracking those who dropped out during the semester and those who dropped out after the semester over several semesters for age and gender variables among the variables that can be collected in the collection unit 120. In addition to age and gender variables, the present applicant tracked and analyzed those who dropped out during the semester and those who dropped out after the semester over several semesters for the various variables already mentioned. Accordingly, we found out which variables would be most desirable to apply to predict dropouts during the semester and dropouts after the semester.

Accordingly, age, gender, entrance grade, quota classification, admission recruitment round, application fee, and basic tuition payment point were used as primary variables, and age, gender, entrance grade, quota classification, admission recruitment round, and application fee were used as secondary variables. , basic tuition payment time, president's scholarship amount, national scholarship amount, course registration credits, and test application rate were used. However, this is only a preferred embodiment, and other variables can be replaced or added.

Figure 4 is a diagram illustrating a method of predicting students scheduled to drop out during a semester through a step-by-step comparison procedure according to the order of variables according to a preferred embodiment of the present invention.

In this embodiment, enroll_credit refers to the course registration credits, scholar refers to the president's scholarship amount, pladytime refers to the content attendance time, out_scholar refers to the national scholarship amount, and age refers to the age.

As shown, the first variable is compared with the corresponding standard level, and according to the comparison result, the next step is performed and the second variable is compared with the corresponding standard level. This process is repeated until the last variable is set, and ultimately, it is possible to predict who will drop out during the semester.

Figure 5 is a diagram illustrating a method of predicting those scheduled to drop out after a semester through a step-by-step comparison procedure according to the order of variables according to a preferred embodiment of the present invention.

In this embodiment, notice refers to the class notice view rate, login refers to the number of logins, ftime refers to the time required to complete attendance, and tution_day refers to the time of basic tuition payment.

As shown, the first variable is compared with the corresponding standard level, and according to the comparison result, the next step is performed and the second variable is compared with the corresponding standard level. This process is repeated until the last variable is set, and finally, it is possible to predict those scheduled to drop out after the semester.

Figure 6 is a flowchart illustrating a method for predicting students who are expected to give up their studies according to a preferred embodiment of the present invention.

The collection unit 120 collects primary variables and secondary variables respectively during a preset collection period (S200). There may be multiple primary variables and secondary variables, and since the relevant information must be collected from the online university's homepage, the collection unit 120 is linked with the network interface unit 110.

When the collection unit 120 completes the collection of the primary and secondary variables, the quantification unit 130 quantifies each of the primary variables and secondary variables using a preset numerical method (S210).

The prediction unit 140 compares the numerical value of the primary variable with a reference level set differently for each variable. As a result of this comparison, if the numerical value of the primary variable is below the standard level (S220-Y), the student is determined to be scheduled to drop out during the semester (S230).

Additionally, the prediction unit 140 compares the numerical value of the secondary variable with a reference level set differently for each variable. As a result of this comparison, if the numerical value of the secondary variable is below the standard level (S240-Y), the student is determined to be scheduled to drop out after the semester (S250).

If, in steps S220 and S240, it is determined that the numerical value of the primary variable or the numerical value of the secondary variable is higher than the reference level set differently for each variable (S220-N, S240-N), the prediction unit 140 ), the student is determined to continue studying (S260).

Figure 7 is a flowchart illustrating a method for predicting students who are expected to give up their studies according to another preferred embodiment of the present invention.

The collection unit 120 collects primary variables and secondary variables respectively during a preset collection period (S200). There are multiple primary and secondary variables, and since the relevant information must be collected from the online university's homepage, the collection unit 120 is linked with the network interface unit 110.

The prediction unit 140 performs an operation of comparing the numerical value of the nth variable among the primary variables with a reference level set differently for each variable. This operation is repeated until all the variables of the preset steps are compared to the first variable (S320).

If the numerical value of the nth variable among the primary variables is below the standard level (S320-Y) and the current comparison step is the last comparison step (S330-Y), the prediction unit 140 will drop the current student during the semester. Decide as a prospective candidate (S340).

Afterwards, the prediction unit 140 performs an operation of comparing the numerical value of the nth variable among the secondary variables with a reference level set differently for each variable. This operation is repeated until all the variables of the preset steps are compared to the secondary variable (S350).

If the numerical value of the nth variable among the secondary variables is below the standard level (S350-Y) and the current comparison step is the last comparison step (S360-Y), the prediction unit 140 will drop the current student after the semester. Decide as a prospective candidate (S340).

If, in steps S320 and S350, the prediction unit 140 determines that the current variable is above the standard level, it determines the current student to be an academic persister (S380).

As a result of tracking and analyzing student-related data repeated several times, we identified variables that tend to exhibit characteristics of students who are expected to give up their studies. In this embodiment, by identifying students' intention to give up their studies before they actually give up their studies based on these variables, various programs can be implemented in conjunction with the students to enable them to continue their studies without giving up.

Those skilled in the art to which the present invention pertains will understand that the present invention can be implemented in other specific forms without changing its technical idea or essential features. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive. The scope of the present invention is indicated by the claims described below rather than the detailed description above, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be construed as being included in the scope of the present invention. do.

100: prediction server 110: network interface unit
120: Collection unit 130: Numerical unit
140: prediction unit 150: storage unit
160: control unit

Claims (7)

Collecting primary variables and secondary variables respectively during a preset collection period;
Quantifying the collected primary and secondary variables;
Comparing the quantified primary variable with a standard level set for each variable to determine those scheduled to drop out during the semester; and
Comprising the quantified secondary variable with the standard level set for each variable to determine those scheduled to drop out after the semester,
The primary variables include age, gender, entrance grade, quota classification, admission recruitment round, application fee, and basic tuition payment time,
The above secondary variables include age, gender, entrance grade, quota classification, admission recruitment round, application fee, basic tuition payment time, president's scholarship amount, national scholarship amount, course registration credits, and test application rate,
The step of determining who is scheduled to drop out during the semester is to compare the primary variable step by step according to the preset variable order, extract students whose numerical values of the variable are below each standard level at each step, and continue until the final step. Students whose numerical values of the variables are below the standard level are judged to be scheduled to fail during the semester,
The step of determining who is scheduled to drop out after the semester is to compare the secondary variables step by step according to the preset variable order, extract students whose numerical values of the variables fall below each standard level at each step, and continue until the final step. A predictive modeling method for a dropout risk group through analysis of learning activity data at a cyber university by a prediction server, characterized in that students whose numerical values of the variables are below the standard level are judged to be scheduled to drop out after the semester. According to claim 1,
A method for predicting a dropout risk group through analysis of learning activity data of a cyber university by a prediction server, further comprising: repeating the collecting step and the quantifying step at each preset collection cycle. According to claim 2,
In the step of determining a student scheduled to drop out during the semester, the repeated performance results are reflected and compared to determine overlapping students as scheduled to drop out during the semester,
The step of determining who is scheduled to drop out after the semester is through analysis of learning activity data of a cyber university by a prediction server, characterized in that students who are overlapping as a result of reflecting and comparing the results of the repeated performance are judged as scheduled to drop out after the semester. Prediction modeling method for dropout risk group. delete delete delete delete

Images