Inferring business strategy by clustering analysis of sentiment in reviews

To open a new business involves many decisions and risks for each one of them. This is a presentation that summarises a report on a study that strives to offer reliable information that may serve as basis for future guidelines regarding a new business. This is done using Yelp's database as the main data source to answer the following questions:

• Which category has a bigger volume of reviews in a city I choose?
• Which city has a bigger volume of reviews in a category I choose?
• Given the last two questions already inferred, what do people most say in reviews using the above chosen category and city filter?

About Las Vegas:

     category review_count star_avg
1 Restaurants       436037 3.418699
2   Nightlife        50012 3.555172

About Restaurants:

       city review_count star_avg
1 Las Vegas       436037 3.418699
2   Phoenix       400892 3.432059
3 Charlotte        69176 3.426986

Sentences wordcloud (left is negative and right is positive):

Through the overviews in slide 3 we can find the biggest volume of reviews, thus answering the first and second questions.

Through the wordclouds in slide 4 we can find what people say in negative and positive points of view.

Both are excellent visualizations that could define a business strategy future. From this wordclouds it's possible to have points of view such as what types of food are most important for customers and what the business should have and in what the business should be better than the competition.

Complete detailed report in: https://www.dropbox.com/s/s7e9y8v4jn2bskz/Report.pdf?dl=0