Flexible, personalized student success modeling for institutions with complex term structures and competency-based education
A flexible persistence modeling system and method for building flexible persistence models for education institutions using a Markov model based on units of academic progress of a non-traditional learning program of an education institution. The Markov model is used to quantify transitions of students between the states as parameters of state transitions so that features from the Markov model with the parameters of state transitions can be extracted that are related to the non-traditional learning program of the education institution using defined flexible persistence. The extracted features can then be used to build at least one flexible persistence model for different segments of the students.
1. A method for building flexible persistence models for education institutions, the method comprising:
translating units of academic progress of a non-traditional learning program of an education institution into states of a Markov model;
instantiating the Markov model to quantify transitions of students between the states as parameters of state transitions;
defining flexible persistence in terms of state-transitional characteristics of the students using the Markov model with the parameters of state transitions, wherein the flexible persistence indicate student enrollment from one collection of academic progress units to another collection of academic progress units;
extracting features from the Markov model with the parameters of state transitions that are related to the non-traditional learning program of the education institution using the defined flexible persistence; and
building at least one flexible persistence model using the extracted features for different segments of the students.
2. The method of claim 1 , wherein translating the units of academic progress of the non-traditional learning program of the education institution into the states of the Markov model includes translating flexible sessions of the non-traditional learning program into states of the Markov model so that each flexible session corresponds to at least one of the states of the Markov model, wherein the flexible sessions are not predefined with respect to sequence.
3. The method of claim 2 , wherein extracting the features from the Markov model with the parameters of state transitions includes extracting features related to the flexible sessions of the non-traditional learning program from the Markov model with the parameters of state transitions.
4. The method of claim 1 , wherein translating the units of academic progress of the non-traditional learning program of the education institution into the states of the Markov model includes translating competency units of the non-traditional learning program into states of a modified Markov model augmented by a hierarchical time-series tree structure so that each competency unit corresponds to at least one of the states of the modified Markov model.
5. The method of claim 4 , wherein the hierarchical time-series tree structure includes parent states representing the students advancing at different speeds for a first competency unit of a course and child states representing the students advancing at different speeds for a second competency unit of the course from the parent states.
6. The method of claim 4 , wherein extracting the features from the Markov model with the parameters of state transitions includes extracting features related to the competency units of the non-traditional learning program from the Markov model with the parameters of state transitions.
7. The method of claim 6 , wherein the features related to the competency units of the non-traditional learning program are based on a sliding window of time so that the features are derived from the sliding windows of time at different times.
8. The method of claim 7 , wherein the features related to the competency units of the non-traditional learning program are based data-adaptive comparison basis so that the features are derived from comparison of the students who progress at a similar rate of competency unit mastery based on the sliding window of time.
9. The method of claim 8 , wherein extracting the features from the Markov model with the parameters of state transitions includes using dynamic time warping for overlapping sessions with respect to time or using anchoring of the overlapping sessions for session comparisons.
10. A non-transitory computer-readable storage medium containing program instructions for method for building flexible persistence models for education institutions, wherein execution of the program instructions by one or more processors of a computer system causes the one or more processors to perform steps comprising:
translating units of academic progress of a non-traditional learning program of an education institution into states of a Markov model;
instantiating the Markov model to quantify transitions of students between the states as parameters of state transitions;
defining flexible persistence in terms of state-transitional characteristics of the students using the Markov model with the parameters of state transitions, wherein the flexible persistence indicate student enrollment from one collection of academic progress units to another collection of academic progress units;
extracting features from the Markov model with the parameters of state transitions that are related to the non-traditional learning program of the education institution using the defined flexible persistence; and
building at least one flexible persistence model using the extracted features for different segments of the students.
11. The computer-readable storage medium of claim 10 , wherein translating the units of academic progress of the non-traditional learning program of the education institution into the states of the Markov model includes translating flexible sessions of the non-traditional learning program into states of the Markov model so that each flexible session corresponds to at least one of the states of the Markov model, wherein the flexible sessions are not predefined with respect to sequence.
12. The computer-readable storage medium of claim 11 , wherein extracting the features from the Markov model with the parameters of state transitions includes extracting features related to the flexible sessions of the non-traditional learning program from the Markov model with the parameters of state transitions.
13. The computer-readable storage medium of claim 10 , wherein translating the units of academic progress of the non-traditional learning program of the education institution into the states of the Markov model includes translating competency units of the non-traditional learning program into states of a modified Markov model augmented by a hierarchical time-series tree structure so that each competency unit corresponds to at least one of the states of the modified Markov model.
14. The computer-readable storage medium of claim 13 , wherein the hierarchical time-series tree structure includes parent states representing the students advancing at different speeds for a first competency unit of a course and child states representing the students advancing at different speeds for a second competency unit of the course from the parent states.
15. The computer-readable storage medium of claim 13 , wherein extracting the features from the Markov model with the parameters of state transitions includes extracting features related to the competency units of the non-traditional learning program from the Markov model with the parameters of state transitions.
16. The computer-readable storage medium of claim 15 , wherein the features related to the competency units of the non-traditional learning program are based on a sliding window of time so that the features are derived from the sliding windows of time at different times.
17. The computer-readable storage medium of claim 16 , wherein the features related to the competency units of the non-traditional learning program are based data-adaptive comparison basis so that the features are derived from comparison of the students who progress at a similar rate of competency unit mastery based on the sliding window of time.
18. The computer-readable storage medium of claim 16 , wherein the features related to the competency units of the non-traditional learning program are based data-adaptive comparison basis so that the features are derived from comparison of the students who progress at a similar rate of competency unit mastery based on the sliding window of time.
19. A flexible persistence modeling system comprising: memory; and a processor configured to:
translate units of academic progress of a non-traditional learning program of an education institution into states of a Markov model;
instantiate the Markov model to quantify transitions of students between the states as parameters of state transitions;
define flexible persistence in terms of state-transitional characteristics of the students using the Markov model with the parameters of state transitions, wherein the flexible persistence indicate student enrollment from one collection of academic progress units to another collection of academic progress units;
extract features from the Markov model with the parameters of state transitions that are related to the non-traditional learning program of the education institution using the defined flexible persistence; and
build at least one flexible persistence model using the extracted features for different segments of the students.
20. The flexible persistence modeling system of claim 19 , wherein the processor is configured to translate flexible sessions of the non-traditional learning program into states of the Markov model so that each flexible session corresponds to at least one of the states of the Markov model, wherein the flexible sessions are not predefined with respect to sequence.
21. The flexible persistence modeling system of claim 20 , wherein the processor is configured to extract features related to the flexible sessions of the non-traditional learning program from the Markov model with the parameters of state transitions.
22. The flexible persistence modeling system of claim 19 , wherein the processor is configured to translate competency units of the non-traditional learning program into states of a modified Markov model augmented by a hierarchical time-series tree structure so that each competency unit corresponds to at least one of the states of the modified Markov model.
23. The flexible persistence modeling system of claim 22 , wherein the processor is configured to extract features related to the competency units of the non-traditional learning program from the Markov model with the parameters of state transitions.
24. The flexible persistence modeling system of claim 23 , wherein the features related to the competency units of the non-traditional learning program are based on a sliding window of time so that the features are derived from the sliding windows of time at different times.