Work in progress entry into the Yelp Dataset Challenge 2016.

The goal for this project is to determine whether the success of a restaurant can be predicted solely on its non-human attributes. For example, does the fact that a restaurant has late hours, serves expensive food, and has a valet imply that it will be more popular? And inversely, does the lacking of certain attributes start a restaurant off on a bad foot?

Using logistic regression, I hope to create a learner that can predict whether a restaurant will fall into one of two categories: Succesful (4 or 5 stars) or Unsuccesful (1 or 2 stars), using only the attributes listed on its Yelp portfolio.

The following is a graph showing the correlation between the number of attributes a restaurant contains, and its likelihood of success:

• Parse for individual attributes in each line
• Using above attribute list, create a list of all unique attributes
• Create a Restaurant by Attribute feature matrix that describes, for each restaurant, which attributes it contains
• Create a binary attribute list for each restaurant to be used with a learner
• Create a binary target list for each restaurant to be used with a learner
• Partition feature vector/target vector into training and test data
• Evaluate accuracy of learner and adjust as necessary
• Implement gradient descent to increase accuracy of learner