Sentiment Analysis Classification

A customer review is feedback provided by a consumer after the purchase and/or use of a business’s product or service. Measuring feedback is important to businesses because it can be a useful gauge for assessing the quality and value that your product or service adds to the customer’s experience. In addition to this, customer feedback serves as a signal for key aspects of business operations such as product development, deciding where to direct ad spend, and building long standing relationships.

In this project, I analyzed an Amazon Alexa dataset consisting of 3,150 samples and 5 original features in order to assess the sentiment of each user’s review. The target feature ‘feedback’ is a binary variable that assigns a value of ‘1’ for positive reviews, and ‘0’ for negative reviews. Using natural language processing (NLP) techniques, I tokenized the ‘verified reviews’ feature in order to transform the dataset into a format that is model training ready. I then utilized a random forest classifier in order to build a machine learning model that can predict the sentiment of given feedback.

The final model reported an overall weighted F1 score of 85%. If you have any questions or comments about the model, please feel free to reach out to me at jasjones82@gmail.com. For further information about the dataset, please visit https://www.kaggle.com/sid321axn/amazon-alexa-reviews. Thank you for reading.