2. Hotel Recommendation System - Inspiration from Zomato Goal: Extract latent (hidden) features from past restaurant reviews and suggest similar restaurants
Description: Rumours are rife on the web. False claims affect people’s perceptions of events and their behaviour, sometimes in harmful ways. Determining what stance majority of people take about a message getting viral on the internet can help to determine the veracity of such rumours. In this task, given a tweet we try to predict if the reply is in Support, Comment, Query or Denial with respect to the source tweet.
Approach: Use the whole conversation context as an input and determine whether the reply tweet is Support, Comment, Query or Denial
Concepts Used:
LSTM - Long Short Term Memory (kind of a neural network)
Hashtag segmentation
Feature engineering
Documentation
Link - For confidential reasons, can't share a link here. Happy to chat with you over my email - [email protected]
Data Visualization
Chicago Crime Analysis
Idea: Analyze the publicly available Chicago Crime dataset to find hidden insights for e.g top crimes, policy arrest activity, etc