Leveraging predictive ML models to improve Automotive Safety and Travel Time through Twitter Sentiment Analysis and Traffic Record Analysis
Authors: Hannah Do, Kamil Sachryn
-
Retrieval of tweets through Snscape, Tweepy, and geocoding through ArcGIS - 1
-
Text processing through various NLP processing and Vader methods (sentiment analysis) - 1
-
Collection of accident record and feature selection - 2
-
Integration of twitter features into traffic accident records - merging the two datasets based on the distance between instances - 3
-
SMOTE and random instance generation for balancing the class - 3
-
ML predictions and evaluation (processed files are in 'merged' folder) - 4
