MITIGATING VARIANCE IN STANDARDIZED TEST ADMINISTRATION USING MACHINE LEARNING

1 . A method for mitigating variability in standardized examination administration, the method comprising:

obtaining, from a database, a first set of testing conditions associated with a first set of standardized examinations;

obtaining, with a user interface, a first set of indications that testing conditions from the first set of testing conditions are irregular;

training an irregularity determination model based on the first set of indications and corresponding testing conditions;

obtaining a second set of testing conditions associated with a second set of standardized examinations;

applying the irregularity determination model to the second set of testing conditions to predict whether any testing condition of the second set of testing conditions is an irregular testing condition; and

reporting the irregular testing condition to the user interface.

2 . The method of claim 1 , further comprising refining the irregularity determination model by verifying, with the user interface, that the irregular testing condition is irregular and modifying the irregularity determination model if the irregular testing condition is not accurately flagged as irregular.

3 . The method of claim 2 , further comprising refining the irregularity determination model by:

calculating a specificity value as a percentage of irregular testing conditions not accurately flagged as irregular as compared with a total number of irregular testing conditions flagged by the irregularity determination model; and

adjusting model parameters for the irregularity determination model if the specificity value falls below a specificity threshold value.

4 . The method of claim 2 , further comprising refining the irregularity determination model by:

calculating a sensitivity value as a percentage of irregular testing conditions flagged as irregular as compared with a total number of irregular testing conditions as verified by the user interface; and

adjusting model parameters for the irregularity determination model if the sensitivity value falls below a specificity threshold value.

5 . The method of claim 1 , wherein the irregularity determination model comprises a machine learning model.

6 . The method of claim 5 , wherein the machine learning model comprises a multinomial naïve Bayes model, a logistic regression model, a convolutional neural network model, or a decision tree model.

7 . The method of claim 6 , further comprising refining the irregularity determination model by:

identifying true positive predictions, true negative predictions, false positive predictions, and false negative predictions from the irregularity determination model;

calculating a confusion matrix from the true positive predictions, the true negative predictions, the false positive predictions, and the false negative predictions; and

adjusting model parameters based on the confusion matrix.

8 . The method of claim 1 , further comprising identifying, with the user interface, disruptive testing conditions that cause disruptions to the administration of the standardized examination and correlating the disruptive testing conditions with irregular testing conditions.

9 . The method of claim 1 , wherein the testing conditions comprise external events, temperature parameters, humidity parameters, seating characteristics, or behavioral characteristics.

10 . A system for removing variability during a standardized examination process, the computer program product comprising:

a user interface;

a database; and

an analytics logical circuit communicatively coupled to the user interface, the analytics logical circuit comprising a processor and a non-transitory memory with computer executable instructions embedded thereon, the computer executable instructions configured to cause the processor to:

obtain, from the database, a first set of testing conditions associated with a first set of standardized examinations;

obtain, from the user interface, a first set of indications that testing conditions from the first set of testing conditions are irregular;

train an irregularity determination model based on the first set of indications and corresponding testing conditions;

obtain a second set of testing conditions associated with a second set of standardized examinations;

apply the irregularity determination model to the second set of testing conditions to predict whether any testing condition of the second set of testing conditions is an irregular testing condition; and

cause the user interface to display the irregular testing condition.

11 . The system of claim 10 , wherein the computer executable instructions are further configured to cause the processor to refine the irregularity determination model by verifying, with the user interface, that the irregular testing condition is irregular and modify the irregularity determination model if the irregular testing condition is not accurately flagged as irregular.

12 . The system of claim 11 , wherein the computer executable instructions are further configured to cause the processor to:

calculate a specificity value as a percentage of irregular testing conditions not accurately flagged as irregular as compared with a total number of irregular testing conditions flagged by the irregularity determination model; and

adjust model parameters for the irregularity determination model if the specificity value falls below a specificity threshold value.

13 . The system of claim 11 , wherein the computer executable instructions are further configured to cause the processor to:

calculate a sensitivity value as a percentage of irregular testing conditions flagged as irregular as compared with a total number of irregular testing conditions as verified by the user interface; and

adjust model parameters for the irregularity determination model if the sensitivity value falls below a specificity threshold value.

14 . The system of claim 10 , wherein the irregularity determination model comprises a machine learning model.

15 . The system of claim 14 , wherein the machine learning model comprises a multinomial naïve Bayes model, a logistic regression model, a convolutional neural network model, or a decision tree model.

16 . The system of claim 15 , wherein the computer executable instructions are further configured to cause the processor to:

identify true positive predictions, true negative predictions, false positive predictions, and false negative predictions from the irregularity determination model;

calculate a confusion matrix from the true positive predictions, the true negative predictions, the false positive predictions, and the false negative predictions; and

adjust model parameters based on the confusion matrix.

17 . The system of claim 10 , wherein the computer executable instructions are further configured to cause the processor to identify, with the user interface, disruptive testing conditions that cause disruptions to the administration of the standardized examination and correlate the disruptive testing conditions with irregular testing conditions.

18 . The system of claim 10 , wherein the testing conditions comprise external events, temperature parameters, humidity parameters, seating characteristics, or behavioral characteristics.

19 . A method for mitigating variability in standardized examination administration, the method comprising:

obtaining, from a database, a first set of testing conditions associated with a first set of standardized examinations;

obtaining, with a user interface, a first set of indications that testing conditions from the first set of testing conditions are irregular;

training an irregularity determination model based on the first set of indications and corresponding testing conditions;

obtaining a second set of testing conditions associated with a second set of standardized examinations;

applying the irregularity determination model to the second set of testing conditions to predict whether any testing condition of the second set of testing conditions is an irregular testing condition;

identifying, with the user interface, true positive predictions, true negative predictions, false positive predictions, and false negative predictions from the irregularity determination model;

calculating a confusion matrix from the true positive predictions, the true negative predictions, the false positive predictions, and the false negative predictions; and

adjusting model parameters based on the confusion matrix;

identifying, with the user interface, disruptive testing conditions that cause disruptions to the administration of the standardized examination and correlating the disruptive testing conditions with irregular testing conditions; and

displaying irregular testing conditions and disruptive testing conditions on the user interface.

20 . The method of claim 19 , wherein the testing conditions comprise external events, temperature parameters, humidity parameters, seating characteristics, or behavioral characteristics.