Exploratory Data Analysis and Sentiment Analysis of Drug Reviews

Abstract

The exponential increase in the volume of data in our daily life needs to be managed and analyzed properly to get knowledge and benefit out of that. A drug review dataset obtained from the UCI machine learning repository has six parameters namely drug-id, name, condition, review, rating, and usefulness count out of which we have filtered a subset of the dataset based on eight conditions. Exploratory Data Analysis (EDA) and Sentiment Analysis (SA) are then applied to the filtered data set. EDA shows the total number of medicines used for all light conditions, number of reviews per condition, five most popular drugs based on usefulness count, number of drugs per condition, etc. SA is performed on the filtered dataset, in which twenty-eight drugs are compared based on rating and polarity where three drugLisdexamfetamine, Vyvanse, Lamotrigine are found to be best in the view of customers as per their rating and positive polarity and Suvorexant is the drug found to have negative polarity and least rating.