z7ye / DS_projects

Data-driven projects repo

Metis Projects Folder

This is my project folder for the metis bootcamp (Spring 2018). You will find 5 main projects in this folder:

01 - MTA Turnstile (Exploratory Data Analysis)

EDA on MTA Turnstile data of NY to find optimal stations to target donors and attendees for a tech event.
Tools: Pandas, Google geocode API
Data Source: NY MTA

02 - Movie Opening Gross Prediction (Regression)

Medium Blogpost
Using Linear Regression to predict the Box Office on opening weekend.
Tools: Web scraping with BeautifulSoup & Selenium, various APIs, Scikit Learn, Statsmodels.
Data Source: Youtube, Google Trends, OMDB (IMDB, Metacritic, Rotten Tomatoes), BoxOfficeMojo, Bureau of Labor Statistics

03 - West Nile Virus Prediction in Chicago (Classification)

Utilized a gradient boosted decision tree to predict the occurrence of a West Nile virus in Chicago using climate & geolocation data. Hosted a Flask app for model prediction and an interactive visualization using Tableau.
Tools: PostgreSQL, CatBoost, Tableau, Flask, D3.js, AWS

04 - Aspect Based Opinion Mining (NLP)

Medium Blogpost
Extracted customers' opinions across different aspects of a restaurant (i.e. food, service, ambience, and price) to derive actionable insights for restaurant owners and to allow Yelp users to compare similar restaurants.
Tools: scikit-learn, MongoDB, word2vec, spaCy, AWS, plotly

05 - Uber Demand Forecasting (Time-series)

Tableau Dashboard
Forecasted hourly demand for Uber across 140 locations in NYC using statistical models and RNN approaches to recommend better locations for drivers and reduce surge pricing events.
Tools: Keras (GRU, LSTM), Tableau, Facebook Prophet