Jason Dolorso
Resume | LinkedIn
Lead Data Scientist at Schneider Electric
Consultant at ACCeSs Lab
Consultant at AIM - IMSG
MSc in Data Science at AIM
A Petroleum Engineer turned Data Scientist with the goal of empowering organizations in making data-driven decisions. Currently a Lead Data Scientist for the world's most sustainable company.
Of the People, By the People, For the People: Predicting US presidential polls per state using Machine Learning¶
Jason Dolorso
Asian Institute of Management
Executive Summary
With the next US presidential election set this November 3, the decision of whether Donald J Trump or Joseph R Biden Jr will sit on the oval office is just around the corner. Contrary to the usual election proces, United States employs an Electoral College process where electors representing each state casts the votes. FiveThirtyEight, a leading poll analysis website, released their national polling averages ahead of the election. There are lots of presidential forecasts being published by different entities but what set FiveThirtyEight's apart is it uses aggregated polling data conducted by other firms and organizations, and also takes COVID-19 into account in their model. This study aims to utilize features such as demographics, unemployment rate, approval rating, and previous election results, to create (1) a classifier to predict the outcome, and (2) a regressor to predict each candidates' percentages, based on FiveThirtyEight's polling averages per state.
Key highlights
- US states are likely to vote the same party that they went for in 2016.
- Current administration’s approval ratings plays a big role in predicting poll results.
- States with lower median age are leaned towards Trump, while higher unemployment rate and asian population are more for Biden.
Full text article and source codes can be provided upon request.