Predict Program Size is a machine learning model that helps you understand the potential size that a current or new program could grow to at your institution. One important note is that all predictions are made in terms of the number of graduates per year at maturity.
In order to estimate program size, there are several possible approaches:
1. Using market data, for example, looking up the Average or Median program completion size for other schools within your market. The limitation with this approach is that these values are not specific to you as an institution.
2. Conducting a peer institution analysis, often by using data from an accrediting body, is another option. The challenge here is selecting peer institutions (those who are similar to you) and not including competing institutions that have different characteristics.
3. Build a machine learning model. This is the approach we have taken with the Predict Program Size app. It allows us to predict program size in a way that is very specific to your institution.
The model looks at both school attributes and program attributes when considering its predictions. School attributes include size, sector, focus & portfolio, selectivity, student types, and location. Program attributes include size, size at similar schools, type, award level, modality, market data, and program portfolio fit.
The following slide highlights the error rate in the model's performance. If a program is actually large, it would only be incorrectly predicted as small 3% of the time. Alternatively, if a program is actually small, the system will predict it to be large only about 2% of the time. These results are from back-testing the model against IPEDS data. The performance of the model was optimized to minimize false positive and false negative predictions as opposed to getting an exact numerical prediction.
Note: The model makes predictions in two forms.
A size category (very small, small, medium, and large) that predicts that your program will be within a specific range. Each size category refers to a size range that may vary by award level. For example, Very Small when predicting bachelors refers to less than 10 (<10) graduates.
A specific most likely size within that range.
By combining the predicted size of a program with data from our Cost Benchmarking Consortium and your inputs, we can now create a detailed Pro Forma Financial Model of a new program launch, allowing you to see the financial impact of a new program.
There are several reasons why you would need an automated and robust pro forma financial tool. These reasons include accelerated insights, risk mitigation, planning, credibility, break-even insights, and stakeholder buy-in.