Comparing Human-labeled and tool-labeled Sentiment

Prepared and presented by: Leah Pope

Blog:on Medium

Introduction

The goal of this experiment is to compare human labeled sentiment to tool labeled sentiment using VADER (Valence Aware Dictionary and sEntiment Reasoner) and Text Blob.

This is a follow-on from my Flatiron Phase 4 Project where I built a text classifer to classify Tweets as either Positive, Negative, or Neutral.

Data

The datasets used in this experiment are from CrowdFlower

• Product Emotions
• Apple Tweets

Repository Structure

--notebooks
----data_cleaning.ipynb
----comparision.ipynb
--data (dir for all raw and processed data files)
--images (dir for images)