In this project, we aimed to demystify descriptions of wine tasting comments, provide useful features/insights to pick high value wines, and recommend wine based on user’s inputs of flavors, tastes, or previous wine tasting experience.

We utilized Sklearn empowered Logistic Regression, Random Forest, and Neural Network models to predict whether a wine would be scored above 90 or not. We scrapped and visualized winery geo locations using Google Geocoding API. We vectorized and performed dimensionality reduction for winery comments. Lastly, we made a wine recommendation system based on cosine similarity of user inputs.

Dataset gatherer from Kaggle post: https://www.kaggle.com/zynicide/wine-reviews