DataMiningProject

Instructions for running algorithms:

PCA

• X -> dataset
• reduceTo -> number of dimensions to reduce

[ XReduced, eigenvals, eigenvecs ] = pca( X, reduceTo )

MLP

• X_train -> training dataset
• expected_train -> training dataset labels
• X_test -> test dataset
• expected_test -> test dataset labels
• X_val -> validation dataset
• expected_val -> validation dataset labels
• num_epochs -> stop criteria, number of epochs
• validation_checks -> stop criteria, number of validations
• num_neurons_hid -> number of neurons in hidden layer
• num_neurons_output -> number of neurons in output layer
• learning_rate -> Gradient Descent learning rate
• momentum -> Gradient Descent with momentum constant
• learning_rate_incr -> factor of increasing adaptive learning rate
• learning_rate_dec -> factor of decreasing adaptive learning rate

[ W1_epochs, W2_epochs, B1_epochs, B2_epochs, mse_train, mse_test, mse_val ] = backpropagation( X_train, expected_train, X_test, expected_test, X_val, expected_val, num_epochs, validation_checks, num_neurons_hid, num_neurons_output, learning_rate, momentum, learning_rate_incr, learning_rate_dec)

DBSCAN

Compile

javac Point.java Dbscan.java

Running

java Dbscan <eps> <minPoints> <path_to_dataset>

• eps -> neighborhood radius
• minPoints -> minimum points to create a core cluster
• path_to_dataset -> path to a txt file separated by \t