VERA - Verifying Remedies Assertions
This project is a machine learning web app designed to assess the evidential basis of unconventional home remedies and combat the proliferation of online health misinformation.
The model was trained on the train_data.csv dataset and is deployed as an interactive web app using Gradio.
About The Project
This repository contains all the code for the project, including:
The Jupyter Notebook used for data cleaning, analysis, and model comparison.
The saved (serialized) machine learning model and vectorizer.
The app.py file that runs the user-facing Gradio web app.
Models Comparison
During the analysis, three different models were trained and evaluated. The model with the best performance (based on AUC and F1-Score) was chosen for the final app.
Model_________________Accuracy__________AUC Score__________F1-Score (Real)
SVM____________________0.9612____________0.9938______________0.9616
LR______________________0.9523____________0.9907__________…
