To classify the potential respective burger chain's customers for McDonalds, JackInTheBox, Sonic DriveIn, Wendy, and ShakeShack by using their tweets for each burger chain.
The datasets of customers' tweets for each burger chain and responses from the chain itself were obtained by Twitter API using Tweepy.
Navies Bayes and Support Vector Machine (SVM) were trained as supervised classification models along Dummy classifier as Baseline model. Target classes were checked for class-imbalance. The accuracy of balanced and orignial datasets were compared to show the improvement of balancing the datasets.
Wordclouds of customer's tweets and responses from the restaruants were displayed to see the most common words.
Topic models with NMF, LSA, and LDA were performed to see the unique clsuters among five different burger chains.
The results were discussed at the end of each section.
The discussion and conclusion were discussed at the end of the notebook.