Machine Learning

Artifical Intelligence and Data Science Projects

ML Algorithms

view code

Python Supervised + Unsupervised Learning 2016 - Present DEVELOPMENT

• A collection of 15 various machine learning algortims implemented from scratch or occasioally using SKLearn.

• Topics include: Classification, Regression, Clustering, Document Similarity, Gradient Descent, PCA, and Streaming.

• Few of my favorite algortims: K-Means, Polynomial Regression, and Bagged Forests.

Wikipedia Analysis

view code

Python Unsupervised Learning 2017 DEVELOPMENT

• Vectorized Wikipedia articles as abstract data types using n-grams, TF-IDF term weighting.

• Clustered articles using Mean Shift clustering to group articles of similar content together in high dimensional space.

• Implemented the PageRank algorithm on 1M articles to create word cloud of the top 100 articles in terms of PageRank.

• Hosted data and ran all algorithms in the cloud using Google Cloud Platform.

Pacman AI

view code

Python Reinforcement Learning 2017 DEVELOPMENT

• Crafted algorithms that teach an artificial agent how to play Pacman.

• Designed tree search algorithms such as Minimax, Alpha-Beta Pruning, and Expectimax to allow the agent to find the most optimal path.

• Utilized a dynamic programming algorithm for solving Bellman-Ford equation in Markov decision process setting to teach the agent the most optimal action in a given situation.

• Wrote reinforcement learning algorithms using value iteration and Q-learning to teach the agent how to maximize it's score

Airbnb Database

view code

Java + SQL Databases 2017 DEVELOPMENT

• Java and SQL project that builds a database that would support an Airbnb type application.

• Includes Java models, SQL queries, and an HTML interface for using the database.

• Some examples of use cases are: registration/login, creating and updating a housing, reservations, visits, favorites, reviews, user trust ratings, housing browse, housing suggestions, and more.

Terroist Info Extraction

view code


• Developed a information extraction system for Latin American terrorism news articles.

• Processes articles and extracts numerous types of information including: type of incident, weapon/s used, perpetrator, perpetrating organization/s, targets, and the vitcim/s.

• Used the Python libraries NLTK and SpaCy to find grammatical patterns in the articles for info extraction.