Recommendations and Simulating Decisions

Key Metrics

Likelihood Score: an estimate of the probability a student will enroll or be retained. 

Recommendations: up to five Othot-derived fields that recommend actions for a specific student to increase their likelihood of enrolling, remaining enrolled, or persisting. These fields can be used as filters and appear in the grid when your models contain prescriptive data, such as financial aid, events, engagement activities, or campus visits.

Creating a What-If Analysis

A What-If analysis simulates the effect of a decision or action on a population of students. It demonstrates what would happen to a population's likelihood scores if a variable were changed. This prediction empowers you to allocate your financial and operational resources more effectively by helping you understand which actions are most impactfull for meeting your goals. To create a What-If analysis:

1. Navigate to the Analyze Page, and click the chevron in the What-If section. 
2. Enter a name and description for the analysis, then click Create new population or Import from Views to define the population.
3. Click Continue.
4. Edit or create your view population. 
5. Choose the variable(s) to simulate your prediction by clicking the check mark. A variable with a lock icon is not available in the analysis you have configured.
6. Click Continue.
7. Review the summary and edit any section as needed.
8. Click Continue. The analysis is running through the Othot data science engine. 
9. View the results by clicking the chevron in line with the report in the grid on the Analyze page.

Understanding a What-If Analysis

The visual chart and the prediction overview table are rich with information about your What-If analysis. A few metrics to highlight in the prediction overview table include:

• Current: the number of records predicted without the What-If changes.
• What-If Impact: the number of additional records predicted with the What-If changes.
• What-If: the new total number of records predicted with the What-If changes.

In the grid, the What-if Score is the new expected likelihood score as a result of the What-If actions that were applied. Delta is a calculated field that subtracts the original Likelihood Score from the new What-If Score. This is a useful metric for determining whether your selected action is likely to have a meaningful impact on each prospect. Sorting the Delta column by descending or ascending order makes it easier to find which prospects are most or least impacted by the What-If scenario.

Creating a Sensitivity Analysis

A Sensitivity Analysis shows how a student’s predicted likelihood changes across a range of values for a single variable. This analysis will run a series of incremental what-if analyses using the Othot prediction models. 

1. Navigate to the Analyze Page, and click the chevron in the Sensitivity section. 
2. Enter the Name and Description for the analysis, and choose to Create new population or Import from View to define the population.
3. Click Continue.
4. Choose a Lifecycle from the drop-down menu.
5. Select sensitivity variables, and choose variables to aggregate the results. 
6. Click Continue. The analysis is running through the Othot data science engine. 
7. View the results by clicking the chevron in line with the report in the grid on the Analyze page.

Understanding a Sensitivity Analysis

The chart provides likelihood scores sorted by the row variable represented. In many cases, you'll want to use the column manager to edit the variables to make the table more digestible. Any variable you choose as an aggregate can be used as a row variable. These variables can be stacked to add more detail to the table. 

To view the delta of the change in the likelihood score, right-click the likelihood field and click Field Settings. Change the format to be an average percentage. 

Creating a Matrix

The Matrix is Othot's optimization engine. It determines the optimal allocation of a variable, most commonly financial aid, across all students simultaneously, within budget and other constraints, to maximize enrollment or minimize aid spend. 

1. Navigate to the Analyze Page, and click the chevron in the Matrix section. 
2. Enter a name and description for the analysis, then click Create new population or Import from Views to define the population.
3. Click Continue.
4. Add or remove filters to further tailor the population, then click Continue.
5. Choose a Lifecycle from the drop-down menu.
6. Choose the type of Matrix:
   • Automatically Populate: selects values automatically that optimize the value per student.
   • Enter Manually: enter values manually, and the model predicts outcomes. 
7. Choose a Template. The template defines the Matrix's objective. The two most common objectives are Maximize Predicted Enrollment and Minimize Predicted Institutional Aid Disbursed.
8. Select Constraints within a template and add the Minimum and Maximum values. 
9. Click Continue.
10. Review and edit your matrix.
11. Click Continue. The analysis is running through the Othot data science engine. 
12. View the results by clicking the chevron in line with the report in the grid on the Analyze page.

Understanding a Matrix

The Objective section displays the optimal result for the goal configured in the matrix template and compares it to the objective's current value. The Matrix outlines how your decision will affect your objective. A few metrics to highlight include:

• Result: the total number predicted with the changes.
• Current: the number of records predicted without the changes.
• Delta: the number of additional records predicted with the changes.

The Constraints section displays the predicted result, current, and delta values for each constraint. The Calculations Section displays the predicted result, current, and delta values for each as calculated by the matrix.

The Matrix Table displays results with all grouping variables as rows. The additional columns are configured based on the chosen template. You can select a variable from the Primary Column drop-down to change it to display as a column. The Secondary Column adds the ability to see the expected enrollment or total record count next to each decision. This is useful for understanding the expected yield rate for each cell in the matrix.

Creating a Scenario

A Scenario combines multiple completed decisions into a single combined view of their total class-level impact. Any completed What-If or Matrix can be applied and stacked to show the predicted outcome if each of the selected tactics were applied.

1. Navigate to the Analyze Page, and click the chevron in the Scenario section. 
2. Enter a name and description for the analysis, then click Create new population or Import from Views to define the population.
3. Click Continue.
4. Add or remove filters to further tailor the population, then click Continue.
5. Drag and drop available reports to select them as part of the Scenario. 
6. If applicable, enter a desired number of top results to be returned. 
7. Click Continue.
8. Review and edit your scenario.
9. Click Continue.
10. The analysis runs through the Othot data science engine. 
11. View the results by clicking the chevron in line with the report in the grid on the Analyze page.

The What-If Score and Delta columns will appear for every record in the grid. Use these metrics to understand which students are predicted to be the most positively and negatively impacted by your potential decisions, including if a Matrix decision was adopted.