Problem

Companies and Corporations have an active flow of costumer opinions through their social media. Clients often access to these sources either to access to information, or to deliver information about their experience.

This information can give the company clues of what can be improved, what may be causing customer unsatisfaction, and what they are doing right.

Nevertheless, to analyze that amount of information can be tedious and require time and resources. Someone would have to read all tweets daily and catalogue them by their sentimental weight.

Solution

This is where Data Science comes in.

Through a larga dataset of tweets from different airlines, it is possible to apply NLP (Natural Language Processing) in order to study what characterizes each different sentiment previously classified. Machine Learning algorithms can learn this patterns and therefore get trained to predict tweet sentiments with certain accuracy (79,36%)

When the algorithm is finally optimized to the highest accuracy results, the maintenance of it will require less resources and time, while increasing its sturdiness over time.

The project report can be downloaded for deeper understanding.