Sentiment Classification System of Twitter Data for US Airline Service Analysis

Abstract

The airline enterprise is a completely aggressive market which has grown hastily within side the beyond 2
decades. Airline organizations motel to conventional purchaser remarks airline enterprise which in flip are very tedious and time consuming. This is where Twitter information serves as an excellent supply to collect purchaser remarks tweets and carry out a sentiment evaluation. In this paper, we labored on a dataset comprising of tweets for six important US Airlines and achieved a multi-elegance sentiment evaluation. This method begins off evolved off with pre-processing strategies used to clean the tweets after which representing those tweets as vectors the use of a deep gaining knowledge of concept (Doc2vec) to do a phrase-degree evaluation. The evaluation changed into finished the use of 7 exceptional class strategies: Decision Tree, Random Forest, SVM, K-Nearest Neighbors, Logistic Regression, Gaussian Naïve Bays and Gadabouts. The classifiers have been skilled the use of 80% of the information and examined the use of the closing. The final results of the check set are the tweet sentiment (positive/negative/neutral). Based at the effects obtained, the accuracies have been calculated to draw an assessment among every class method and the average sentiment rely changed into visualized combining all six airlines.