# 4.3.13 Healthcare Costs - Video 7: Baseline Method and Penalty Matrix | Summary and Q&A

1.5K views
December 13, 2018
by
MIT OpenCourseWare
4.3.13 Healthcare Costs - Video 7: Baseline Method and Penalty Matrix

## TL;DR

The baseline method for predicting healthcare costs in 2009 based on costs in 2008 had an accuracy of 68% and a penalty error of 0.74.

## Install to Summarize YouTube Videos and Get Transcripts

### Q: What is the accuracy of the baseline method for predicting healthcare costs in 2009?

The accuracy of the baseline method is 0.68, meaning it correctly classified 68% of observations in the test set based on their cost buckets in 2008.

### Q: How is the penalty error computed for the baseline method?

The penalty error is computed by multiplying the classification matrix by a penalty matrix, which assigns penalties for predicting a different cost bucket than the actual outcome. The penalty error for the baseline method is 0.74.

### Q: What is the goal for the upcoming CART model?

The goal for the next video is to create a CART model that outperforms the baseline method by achieving a higher accuracy than 68% and a lower penalty error than 0.74 in predicting healthcare costs.

## Summary & Key Takeaways

• The baseline method predicts that the cost bucket for a patient in 2009 will be the same as it was in 2008.

• An accuracy classification matrix is created to measure the accuracy of the baseline method on a test set.

• A penalty matrix is created to compute the penalty error of the baseline method by multiplying the classification matrix by the penalty matrix.