Problem: In this problem you will use hold-out validation (80% 20%) to select number of cluster centers for an RBF classifier model. You will be using "healthcareTrain.csv" and "healthcareTest.csv" data sets to predict the pdc-80-flag using the following continuous features: total-los, num-op, num-er, num-ndc, pre- total-cost, and pre-CCI.
1. Use rbf function in R to create your RBF model. Try cluster numbers 5 to 25. Use your hold-out validation set to figure out the best number of clusters. Report the best number of clusters and the corresponding validation accuracy rate.
2. Plot the accuracy rate vs. number of cluster centers.
3. Use the model with the best performance to predict the pdc-80-flag for the test set. How does your validation accuracy rate compare to test accuracy rate?
Attachment:- Data.zip