In this opportunity we'll go through the analysis of the dataset available at https://www.kaggle.com/thoughtvector/customer-support-on-twitter/ which contains several tweets from the customer support service of multiple companies around the globe. Essentially we're going to be focused on how the service is, based on the tweets obviously, how they can be improved in a business scenario and how the clients express themselves about the service. At the end, we'll perform topic modeling to find out what are the hottest topics the clients request assistance for.
We won't go through classification, only topic modeling. The reason is because we only want to explore the service and provide some conclusions together with possible solutions to improve the overall support and not to develop solutions, just infer them. We'll get to the point where, if the business would require it, the data would be ready for predicting automated responses (maybe implementing chatbots) or just provide links with how-to articles to increase the self-resolution.
We'll show how the dataset is composed and then will deep dive into it from the most general to the most specific.