The goal of this problem is to compare the results of the analysis of the basketball data obtained by projection pursuit to the results one can obtain using neural network regression instead.
1. Produce numerical evidence that the variable age does not add to the prediction of the variable points and/or the variable assists in the case of neural network regression.
2. Use the same training sample BT RG to fit a neural network and use the coefficients (i.e. weights) of the model to predict the values of the variable points in the test sample BTST. Compute the residual sum of squares for these 10 predictions and use this value to say which of the projection pursuit or the neural network method did better (for this particular criterion and on this particular data set).