This project requires this sklearn library. The project involves taking a training dataset consisting of 30,000 features and extracting at most 15 features which are the most significant (this was done by taking the pearson coefficient) and then running a custom alorigthm. The steps are as follows: i) Read data and labels ii) Overall Feature selection (Reduce from approx 30k to 2k) iii) 5-fold cross validation: a)Pearson coefficients calculated b)Four classifiers used (worth 1 "count" each) c)Accuracy depends on the sum of the value counts by each classifier: ex. svm predicts 0, logistic_regression predicts 0 gaussian_nearest_means predicts 0 and nearest_centroid predicts 1 value = 0 + 0 + 0 + 1 if value <= 1 then classify in 0 if value >= 3 then classify in 1 else (if value = 2 or other) then classify as svm predicted iv) Read test data and perform feature selection (features from train data) [extract 15 columns] v) Output the num of features and the features themselves on console & save test labels as a file named "sh486_testLabels" This was a project I made for course CS675 [Machine Learning] at NJIT Fall Semester 2017.

Python

MachineLearningProject

The steps are as follows:

i) Read data and labels ii) Overall Feature selection (Reduce from approx 30k to 2k) iii) 5-fold cross validation: a)Pearson coefficients calculated b)Four classifiers used (worth 1 "count" each) c)Accuracy depends on the sum of the value counts by each classifier: ex. svm predicts 0, logistic_regression predicts 0 gaussian_nearest_means predicts 0 and nearest_centroid predicts 1

             value = 0 + 0 + 0 + 1
             if value <= 1 then classify in 0
             if value >= 3 then classify in 1
             else (if value = 2 or other) then classify as svm predicted

iv) Read test data and perform feature selection (features from train data) [extract 15 columns] v) Output the num of features and the features themselves on console & save test labels as a file named "sh486_testLabels"

This was a project I made for course CS675 [Machine Learning] at NJIT Fall Semester 2017.

koalasession/MachineLearningProject

MachineLearningProject