Here are some of my projects:
Analyzing Rideshare Prices
When you use a platform like Uber or Lyft, we're usually led to believe that the distance, time, and type (SUV, XL) of ride directly affect the price one has to pay. Through data exploration and analysis of a rideshare dataset, my group was able to discover whether other factors would affect a rideshare's price, like weather or temperature. Ultimately, we built a multiple linear regression model to predict a rideshare price based on a variety of factors: including those weather-related. Furthermore, we conducted various hypothesis tests and utilized hyperparemeter tuning to optimize our model(s). Check out more here!
Neural Networks and Sentence Generation
In this project from my Natural Language Processing class, my partner and I utilized neural networks to generate new sentences from a training corpora. In order to do this, we generated word embeddings from preprocessed text using the skip-gram model. Then, we trained these embeddings on a keras feedforward neural language model. Lastly, we used our trained models to generate sentences with Shannon's method. We used two datasets: one was the Spooky Authors Dataset and a lyrics dataset from popular artists. If you are interested in how we did this, or what sentences we generated, check out more here!
Sentiment Analysis in Video Game Reviews
I love video games and machine learning, so why not combine them both into a project? I decided to use Python to analyze a set of video game reviews. By vectorizing the text data, I was able to feed this information into various models and predict the sentiment (Positive or Negative) of a given review. I also used hyperparemeter tuning to slightly improve performance of my models. In my NLP class, I built on this by adding non-bag-of-words data such as sentence length and puncutation counts to my data. In addition, I used other models such as RFC and K-Neighbors Classifier to tackle this sentiment analysis task. In the future, I might try dabbling with neural networks to compare their performance with my previous models. Check out more here!
ServiceNow Co-Op: User Journey Analysis
During my co-op at ServiceNow, I pioneered the user journey analysis for the MarTech Data & Analytics team. This project consisted of extracting user web behavior, form submissions, and sales milestones data, and then stitching it all together into a single dataset. After cleaning, normalizing, and wranngling the data, I performed an exploratory analysis on this dataset, highlighting insightful trends that may prove to be useful to the sales team(s). Ultimately, I noted data deficiencies in my created dataset and planned to build predictive models for the rest of my team to address. I am proud of the work I accomplished in several months, and maybe I'll be able to return and build upon it in the future!