State Farm Classifier

Inspiration

Our Inspiration was the State Farm's slogan which is "Like a good neighbor, State Farm is there." From that point we decided to work on increasing customer satisfaction and what could we do different then other insurance companies what would make State Farm customers more special.

What it does

The main goal was to be sure every State Farm Customer feels they are heard and State Farm actually care about their reviews and feedbacks. What we did was a filtration system that will sort every feedback received and sort them by insurance type, policy name and either it is positive or negative. Which will show the things State Farm needs to work on.

How we built it

The program uses object oriented programming to split the program across specialized classes.

The sentiment analyzer class gets text, formats it, cleans it up, and then returns a sentiment score of the cleaned text_. The result is formatted and returned as a percentage.

The Review Instance class is to create review objects that hold all the data that is inputted. It has the customer's name, date of review, review, sentiment value, member duration, current policy, and policy type. Each individual attribute of an object can be easily pulled or edited.

the CSVreader reads in reviews from a datafile storing them as review objects in a list. It returns this list to be sorted and outputted by main.

Main connects all the classes and outputs the sorted feedback for users.

Challenges we ran into

It was a long process to decide how to sort the data and the hierarchy level of importance with it.

Accomplishments that we're proud of

We used various file processing techniques as well as some natural language processing to compile, sort, and judge various reviews based on their content and their relationship to other reviews

What we learned

We learned how to edit the data we have at the hand and how to create a system beneficial for company and customers at the same time.

What's next for State Farm Classifier

We had so many ideas of how to expand on this program, including a) including an easy-to-use interface for both clients and State Farm developers, and b) notifying customers when their issue has been fixed and what they think of the improved product. We also hope to add more subcategories of policy types for pinpoint accuracy.