Pattern Recognition
Spring 2018 - Project
M. Apostolidou and N. Kyparissas

Text Categorization using Gaussian Mixture Models

-----------------------------------------------------------------------------
Contents:
-----------------------------------------------------------------------------
data 				: contains training and testing data set
results 			: contains csv files with accuracy results
figs				: contains figures from performance analysis
docs				: contains the project report and presentation slides
code.py				: main code file, to perform training and testing
dictionary_sort.py	: methods for the initial feature reduction 
tfidf.py			: methods to create tf-idf matrices
datAnalysis.py		: code to analyze csv results in pandas
run_stats.sh		: bash script to run code.py and collect accuracy results


To collect results for parameters 

initial feature reduction: 0 500 1000 1500 2000 2500 3000
svd components: 15 20 25 30 35 40 45 50
gmm components: 1-25

run run_stats.sh

To train and test the code 
run code.py num_initial_feature_reduction num_svd_components num_gmm_components