Basic Spelling Checker

    Implemented a basic spell checker inspired by Peter Norvig (https://github.com/soutik/Spell-Checker)

  • Used data from Peter Norvig 'big.txt' to create a repository of word frequencies
  • Modified input word to obtain several variations of it by operations like delete, insert, swap and add new character
  • Achieved 4% more accuracy by modifying Peter Norvig's implementation

KNN - Boston Housing Dataset

    In this project we implement the KNN algorithm from scratch and use it on the Boston Housing Dataset

  • Performed exploratory data analysis.
  • Implemented N-Nearest Neighbours
  • Implemented KNN algorithm

NYC Flight Analysis

    In this problem set I used the data on all flights that departed NYC (i.e. JFK, LGA or EWR) in 2013. One can find this data as part of the 'nycflights13' R package. Data includes not only information about flights, but also data about planes, airports, weather, and airlines.

  • Performed exploratory data analysis to find what are the seasonal and daily trends in flight delays.
  • Used co-relation matrices and p-value test to figure what variables actually affect the flight performance index/delay across the year 2013
  • Suggested reasons for the pattern discovered during this process

NYC School SAT score

    Analysis of trends in SAT scores across all schools in NYC. Finding trends in different sections of the SAT exam and figuring the top 5 schools in NYC to increase the probability of getting great SAT scores.

    Used R to extract JSON data of all school’s SAT data from https://nycopendata.socrata.com website

  • Created linear model to show trends between different sections of the SAT test in each school
  • Analyzed the top 5 schools to be in NYC for a great SAT score.

Twitter Sentiment Analysis
  • Used TwitteR package inside Twitter to pull tweets containing a certain word/phrase
  • Created 'Sentiment' function inside R to analyse each tweet and attaching an emotion that the tweet showcases.
  • Created a wordcloud including the most frequent words found in those tweets.
  • This project can be used to find any word and any number of tweets (As permitted by the Twitter API by just changing 2 words in the code).