Skip to content

bhanuvikasr/Text-Mining-Yelp

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Text-Mining-Yelp

Abstract

In this project we investigate a variety of regression and classification models to predict rating from given review text. It was found that Ridge regression with TFIDF unigram features worked the best in terms of Mean Squared Error. We investigate the model performance on data from different cities and conclude that our model is robust to changes in rating distribution. Also, we show that the text based models should be used with caution across regions since they are sensitive to language. Additionally, it is observed that it may be better to use different models for different time periods (seasons or years). Visualizing the word clouds give valuable insights about region, culture as well as its correlation to seasons and ratings.

Dataset

Round 9 Of The Yelp Dataset Challenge - link
Below is the preview of relevant data files.

yelp_academic_dataset_business.json

{
    "business_id":"encrypted business id",
    "name":"business name",
    "neighborhood":"hood name",
    "address":"full address",
    "city":"city",
    "state":"state -- if applicable --",
    "postal code":"postal code",
    "latitude":latitude,
    "longitude":longitude,
    "stars":star rating, rounded to half-stars,
    "review_count":number of reviews,
    "is_open":0/1 (closed/open),
    "attributes":["an array of strings: each array element is an attribute"],
    "categories":["an array of strings of business categories"],
    "hours":["an array of strings of business hours"],
    "type": "business"
}

yelp_academic_dataset_review.json

{
    "review_id":"encrypted review id",
    "user_id":"encrypted user id",
    "business_id":"encrypted business id",
    "stars":star rating, rounded to half-stars,
    "date":"date formatted like 2009-12-19",
    "text":"review text",
    "useful":number of useful votes received,
    "funny":number of funny votes received,
    "cool": number of cool review votes received,
    "type": "review"
}

Tools used

  • Numpy, Scipy, Pandas
  • Sklearn, NLTK
  • Matplotlib, WordCloud

Releases

No releases published

Packages

No packages published