A Statistical approach to evaluate the efficiency and effectiveness of the Machine Learning algorithms analyzing Sentiments

In the process of analyzing sentiments for a given dataset various machine learning techniques are used. The models using these learning algorithms help in determining the sentiments across the textual documents. There is a need to evaluate the effectiveness of the models in terms of analyzing and predicting sentiments. This paper provides a statistical approach to measure the effectiveness of the models and also evaluates their effectiveness with respect to the data representations. Here an experimental research is carried out with an inductive mode to measure and evaluate the models. The models are built using Decision Tree, Naive Bayes and Support Vector Machines. Data has been represented using features of Term Frequency and Inverse Document Frequency and Bag-of-words. Statistical tools used for measuring the models are Chi-square test and Analysis of Variance.