Skip to content

Deep neural network models to classify various drug categories from online drug reviews

Notifications You must be signed in to change notification settings

tkhang1999/drug-category-classification

Repository files navigation

Drug Category Classification

Introduction

This projects aims to build two deep neural network models, namely CNN and Bi-LSTM, to classify drug categories from a set of online drug reviews using Tensorflow/Keras and UCI Drug Review dataset.

Our models are able to classify 14 different drug categories from drug reviews with an overall accuracy of approximately 89%. An online version of this notebook is available on Colab.

How to run?

  • Require python 3.8 or higher

  • Install all dependencies for this project

$ pip install -r requirements.txt
  • Start Jupyter Notebook
$ jupyter notebook

Experiment results

  • CNN model with a Word Embedding layer using GloVe achives an accuracy of 89.24%

CNN result

  • Bi-LSTM model with a Word Embedding layer using GloVe achives an accuracy of 88.91%

LSTM result

About

Deep neural network models to classify various drug categories from online drug reviews

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published