13.6 Exercises

  1. Repeat the vowel classification example on PCA-reduced data.

  2. Conduct a pre-analysis exploration as in the Wine example to remove variables in the 2011 Gapminder, the Iowa Housing, and the Vowel datasets before conducting the analyses, as in the examples.

  3. Construct and evaluate naïve Bayes classifiers for the Wine and for the 2011 Gapminder dataset.

  4. Construct and evaluate CART models for the Wine and for the Wisconsin Breast Cancer datasets.

  5. Construct and evaluate ANN models for the 2011 Gapminder, for the Iowa Housing, for the Vowel, and for the Wisconsin Breast Cancer datasets.

  6. Re-run the ANN models incorporating 10 hidden layers with 30 nodes. How much more time does it take to run a “bigger” neural network on the Wine dataset?

  7. Build bagging models for the 2011 Gapminder, Wisconsin Breast Cancer, and Wine datasets.

  8. Build random forest models for the 2011 Gapminder, Wisconsin Breast Cancer, and Iowa Housing datasets.

  9. Build boosted models for the 2011 Gapminder, Wisconsin Breast Cancer, Wine, and Iowa Housing datasets.

  10. Build classification models for the datasets GlobalCitiesPBI.csv, 2016collisionsfinal.csv, polls_us_election_2016.csv, UniversalBank.csv, and HR_2016_Census_simple.csv and/or any other datasets of interest. You may need to identify/define a categorical response variable first.