Automated Classification of Depressive Tweets Detection Using Machine Learning and Deep Learning Techniques

Main Article Content

Mohhammed Jasmine , Sandeep Yelisetti

Abstract

The most occurring emotional mental disorder now a days is Anxiety and depression. Where anxiety and
depression are related and directly proportional to each other. Anxiety and Depression are two very dangerous and harmful diseases or disorders which occurs in children, undergraduates, teenagers, employees and old people. More anxiety and depression have suicidal tendency. Depression occurs due to many parameters like tensions, lack of sleep, lack of confidence and etc. The person who suffers from depression have the higher probability to suffer with anxiety. The main problem with these disorders is that detecting anxiety and depression is a very difficult task, therefore one of the way to detect the people who are suffering from the anxiety and depression is with the help of social support by using the twitter data and extracting the data from the tweets which they post and analysing the data and detecting whether the tweets posted by the people on twitter are depressive or not by using best machine learning techniques like naïve Bayes, logistic regression and also the deep learning techniques like CNN and LSTM. The naive Bayes classifier is primarily used in sentiment classification, and it is based on the Bayesian network. The Naive Bayes method predicts text based on the frequency of words in the text. In contrast, logistic regression analysis is a statistical probabilistic classifier that is used in deep learning to estimate the probability of a binary response based on one or more predictors. With the aid of social media (twitter) data, the final result of this research is to classify whether a given tweet from the dataset is depressive or not, and applying feature extraction methods to this classifier helps to improve the accuracy results of the classifier. The TF-IDF feature extraction method was used in this study (Term frequency and inverse document frequency). And whereas in deep learning, LSTM and CNN are used to predict the depressive tweets with best accuracy.

Article Details

Section
Articles