Pratt Singh

Hi, I’m Pratt! I graduated from the University of Washington-Seattle and currently work as an Analyst at Starbucks. As part of the Pricing Strategy and Applied Economics team, I have developed tools and models which help drive business impact. The intersection of business and data oriented decisions has always been my passion, and I love collaborating with diverse stakeholders that this field brings!

Outside of work, I love watching movies, hiking, trying new cafés, and cuddling with my cat!

Portfolio // Projects

Deploying Machine Learning Models with Django

Tools used: Machine Learning with Python (Sklearn, pandas, numpy), Django (REST, POST), SQLite

The repository can be viewed by clicking here

Update 2023: In light of the situation in recent years, I have decided that this project is insensitive to the reality of systematic oppression that has existed in this country. As users of data, we should always practice and recognize the ethical boundaries of our analyses. When I did this project in 2020, I was taking it at face value, but now I think it is important that we share the message of responsible data use.
2020: The purpose of this project was to build a Machine Learning model that predicts income level based on census data, and then to deploy that model to a server using Django. In this project I utlized the Random Forest and Extra Trees algorithms for machine learning predictions, and built a django server that had interactivity using REST API. SQLite was the database of choice. Finally, I ran an A/B test to see which algorithm (random forest vs extra trees) yielded better results.

Finding the Best Markets to Advertise an E-Learning Product

Tools used: Python (Pandas, NumPy, matplotlib, seaborn).

Suppose that a company offered an E-Learning subscription based product that taught users to code. Which markets do we advertise in to maximize outreach and growth? In this project, I decided to analyze how a product offering coding courses can target markets based on data from a survey. I cleaned the data, visualized, corrected for outliers, analyzed and then concluded with the 2 best markets to advertise in. The notebook and code can be viewed by clicking here.

Exploring Diversity in The Academy Awards (Oscars)

Tools used: Python (NumPy, Pandas, matplotlib, seaborn), Tableau, Excel.

The Oscars are the ultimate recognition of merit in the film industry, and are won by a few. These awards are often the crowning achievement for actors, and greatly compliment their career and image. The Academy Awards have been given since the 1920s, making them nearly a 100 years old.

I decided to analyze the demographics of the Academy Award winners, to deduce how the diversity of winners has evolved in the last 100 years. The notebook and code can be viewed by clicking here.

From the analysis performed in Python, I created and joined a new dataset which I used to create a visualization dashboard in Tableau. It can be viewed and interacted with below. The dashboard showcases the historical trends in diversity + representation in the Academy Awards. While more POC are being awarded, the Oscars have remained overwhelmingly cis-white. A few possible explainations exist. Firstly, due to the economic disadvantage that POCs face, arts and acting are less likely to be the primary economic motivators for them. Secondly, movie directors and their biases towards white characters can impact the number of roles suitable for POC. UCLA’s Hollywood Diversity Reports show that while POC make progress, they are dramatically underrepresented.

CONTACT

email: singhp98@uw.edu or prat.singh2016@gmail.com

LinkedIn: linkedin.com/in/prattsingh/