• Universal Basic Data Income
• Application that monetizes personal

data, while keeping it completely anonymous.

​

​

Gather survey results to analyze and compare changes in moods of political parties/figures during political events.

​

​

​

Thank you to our professors, GSIs, and UBDI for their guidance and encouragement!

UBDI, Inc. Sentiment Analysis Tool

Erin Neaton, Sami Herzog, Sean Yoon, and Thomas Bautista

What is UBDI?

Goal Statement

Project Overview

Visualizations/Results

UBDI Tweet/ Political Tags CSV

+

Trump Tweet CSV

Data Cleaning

Vader Sentiment Scores

Pandas Data Manipulation

Bokeh and Seaborn Visualizations

Acknowledgements

UBDI, Inc. Sentiment Tool

Sami Herzog, Thomas Bautista, Sean Yoon, Erin Neaton

UBDI: Universal Basic Data Income

• App that compensates users for anonymous, personal data
• Conducts market research to sell to clients

Goal Statement

Gather survey results to analyze and compare changes in moods of political parties/figures during political events.

Process/Project Overview

UBDI Tweet/ political tag Data

+

Trump Tweet CSV

Clean data using natural language processing techniques

Infer sentiment scores using Vader Sentiment library

Reformat and manipulate data using Pandas library

Visualize sentiment patterns using Bokeh and Seaborn libraries

Data Collection

• Marketed a survey on the UBDI application to collect historic Twitter post data
• Survey resulted in millions of rows of data including Tweets and users’ political parties
• Augmented with CSV of Trump’s Tweets to serve as comparison to a political figure
• Cleaned raw tweets and reformatted the datasets into one

Sentiment Analysis

• Utilized Vader Sentiment library in Python to assign sentiment score to each Tweet
• Vader is specialized in assigning sentiment to Tweets and is well-established in academia
• Integrated both Seaborn and Bokeh libraries and its visualizations to communicate trends
• Allow for interactive visualizations

Trump’s and Political Parties’ Average Sentiment about Trump

Trump’s and Political Parties’ Sentiment about Trump in the 2016 Election

Insights from Visualizations

• Overall, Trump tweets positively about himself while democrats tweet more negatively about him. Republicans tend to fall in the middle.
• Trump’s average sentiment about himself on election day was very positive.
• Democrats’ average sentiment about Trump was very negative on election day and in the following days.
• Republicans were moderately positive in sentiment about Trump following his election.
• No other parties seemed to tweet about Trump in this time.

​

Value to UBDI

• Businesses can better understand their consumers by analyzing their moods about products during different periods of time, such as product releases or announcements
• Value of UBDI’s service and network effect increases

​

Trump’s and Political Parties’ Average Sentiment about Trump

Trump’s and Political Parties’ Sentiment about Trump in the 2016 Election

HeatMap of Political Parties’ Sentiment

• Analyzed how the moods of UBDI consumers change during different time of the year, holidays, seaons, etc.

​

Political Parties’ Average Sentiment Score Across Time (HeatMap)

Political Parties’ Average Sentiment Score Across Time (Interactive HeatMap)