Hakizimana, Lyse Deborah <br/><br/>
EVALUATING CUSTOMER SATISFACTION USING  MACHINE LEARNING TECHNIQUES / 
 CASE STUDY  OF U.S. AIRLINE INDUSTRY 
LYSE DEBORAH HAKIZIMANA; SUPERVISOR: ASST. PROF. DR KIAN JAZAYERI 
 - vii, 51 sheets;  31 cm.  1 CD-ROM 
<br/><br/>Thesis (MSc) - Cyprus International University. Institute of Graduate Studies and Research Management Information Systems Department 
<br/><br/>Includes bibliography (sheets 50-51) 
<br/><br/> ABSTRACT<br/>In today's industry environment, the airline company is increasingly focused on <br/>comprehending and responding to client needs and feedback to improve general <br/>levels of satisfaction. By measuring tweets sentiment, airlines can win a complete <br/>comprehension of customer‟s sentiments and opinions which can help airline <br/>industries in the creation of quality improvement plans. In this study, we used <br/>sentiment analysis, one of which understands text-sensitive applications, to find <br/>consumers‟ opinions. Machine learning algorithms have been developed for <br/>categorizing Twitter postings as positive, negative, or neutral. We trained our model <br/>on tweets from six US airlines, and this study explores the use of predictive analytics <br/>using the Naive Bayes algorithm which is a useful tool for extracting useful insights. <br/>The text data is cleaned using preprocessing techniques to ensure adequate input for <br/>the Naive Bayes classifier. Random Forest classifier and the K-Nearest Neighbors <br/>algorithm were combined with a model constraint classifier to increase the accuracy <br/>of Naive Bayes after selecting the most essential features using SelectKBest. This <br/>study examines the accuracy of Naive Bayes used for sentiment classification in <br/>Twitter data. The findings and insights can be utilized to discover effective practices <br/>as well as areas for more investment in order to improve customer satisfaction. The <br/>algorithm's probabilistic approach can classify tweets in negative, neutral, or positive <br/>sentiments by allowing airlines to identify areas of customer complaint and areas of <br/>excellence. In this work, the performance parameters of the proposed solution were <br/>considered to be recall, precision, and F1-score. Naive Bayes classifier shows a <br/>capacity to correctly categorize the opinions conveyed through tweets. The collected <br/>findings offer light on common consumer complaints, allowing airlines to promptly <br/>resolve issues and target initiatives to increase success.<br/>Keywords: Customer satisfaction, K-Nearest Neighbors, Naive Bayes, Random <br/>Forest, Select KBest, Sentiment Analysis, Stacking Model, Twitter Data and US <br/>Airline 
<br/><br/><label>Subjects--Topical Terms: </label><br/>Consumer satisfaction--Dissertations, Academic<br/>Sentiment analysis--Dissertations, Academic