ruBayes

copyright Anshuman Chhabra (2016)

• ruBayes is a script that allows you to perform multi class classification on most datasets (usually, smaller work better). The ones I have used here come from the UCL Machine Learning Repository

Dependencies

• ruBayes uses the descriptive_statistics rubygem.
• install it using gem install descriptive_statistics in terminal.

Usage

• Run ruBayes.rb in terminal using ruby ruBayes.rb

• You will now have to enter the names of the classes for classification. For example, in test2.csv and train2.csv the classification is made between Iris-setosa, Iris-versicolor and Iris-virginica. So you will enter Iris-setosa Iris-versicolor Iris-virginica at the prompt

• The code will output an accuracy based on the number of classifications it got correct. For the tested data for both sets, ruBayes got above 92% accuracy.

• The datasets for training and testing are placed in the train and test folders respectively

• The data needs to be formatted in a particular way. Although, most UCI ML datasets are formatted this way by default so just downloading a dataset from there and going ahead with the script shouldn't be a problem.

• To add your own test and train csv files just replace them in their respective folders. Make sure if you change the name from test.csv/train.csv or test2.csv/train2.csv you also make the requisite changes to the preprocess.rb file. It basically processes the files and the data in them and feeds it to ruBayes.

Links

• The datasets used are Haberman's survival (dataset1) and Iris (dataset2).