Dimensionality-Reduction-and-Single-Multi-layer-perceptrons

Data:

Iris data set: 4 inputs and 5th value is the class code (1,2,3). There are a total of 150 instances, 50 per each class; use 30 of each class for training and 20 for testing.

Classifiers:

1. Linear perceptron with 4 inputs and 3 outputs for classification
2. Multilayer perceptron (MLP) with 4 inputs, 2 hidden units and 3 outputs for classification
3. Use PCA to reduce dimensionality to 2 and then use a linear perceptron with these 2 inputs and 3 outputs
4. USe LDA to reduce dimensionality to 2 and then use a linear perceptron with these 2 inputs and 3 outputs

Performance Assesment:

For these 4 models, calculate confusion matrices on training and testing data. For 2,3,4, plot the training and test data (in two seperate plots) in the 2D sapce of hidden units for MLP