and Schizophrenia), what are the main textual features that can improve the understanding of user satisfaction for certain drugs?
This project analyzes psychic drug reviews using sentiment analysis to better understand textual predictors for patient satisfaction. Drug reviews were used to train a Naive Bayes and Random Forest classifier. Furthermore, the most important features were visualized with the aim to improve the understanding of different textual features for five psychic diseases (ADHD, Anxiety, Bipolar disorder, Depression, and Schizophrenia).
The drug reviews were sourced from scrapdrugs on GitHub, a curated version of the dataset from Drugs.com. The datasets for the five mental disorders studied are provided as follows:
The code for the analysis is available in the repository under Project Code. The report containing findings and methodology can be accessed here: Project Report.
The groundwork for this research, including the selection of the research topic and the initial stages of data analysis, was a joint effort between myself, Ann-Kristin Balve, and my project teammates: Anita Braida, Christophe Friezas Gonçalves, Ivonne van der Heiden, and Sally Soeng. We worked collectively to establish the foundation upon which this research was built.
This project is licensed under the MIT License - see the LICENSE file for details.