Both of the two problems are about classification. You are expected to write a report consisting of
1) A description of the datasets (10%)
2) Description of data preprocessing, how the data is split as training and testing set (20%)
3) A conclusion or summary of results (40%). The reported accuracy needs to be averaged over 100 runs.
4) Include your R code as appendix to the report (20%)
5) Clarity of the report carries 10%.
6) Submit in class by hard-copy.
1. Please find a dataset for which neural network does very badly, comparing to logistic regression.
2. Please find another dataset for which neural network outperforms logistic regression. For this dataset, change the number of hidden units (nodes) until the performance of neural network deteriorates. Plot the accuracy as a function of the number of hidden units.