Project description

4P76

The main aim of the project is

• To determine which of the ML methods discussed best classifies unseen data for a given dataset.

The tasks are as follows:

1. Describe the data set.
2. Choose at least two methods from each of the statistical and rule based groups and one neural network algorithm.

   Statistical	Rule based	ANN
   Discriminants	ID3, C4.5, C5	Perceptrons
   k-nearest neighbor	CART	Radial basis
   Naive Bayes	Bayes tree	DIPOL92
   Causal networks	Rough sets

3. Describe the chosen algorithms, and discuss their advantages and disadvantages.
4. Develop a setup for testing the prediction quality of each method. This includes software selection and a brief description of the software.
5. Validate your results by using the validation methods given below. Describe these methods and their pros and cons.

   Validation methods

   Jack-knife

   Cross validation

   Bootstrap

6. Compare the performance of the algorithms in terms of prediction quality, complexity, storage costs etc.