Every wine has numerous features, from its country of origin down to the vineyard it hails from, the awards its won to the scores its received. Pricing wines therefore can sometimes seem arbitrary in light of similar origins and reviews, yet yielding wildly different costs. Applying a machine learning solution to this question seems natural, but due to the lack of numerical data it will require some feature engineering.
All of this data was scraped by me from https://wine-searcher.com
- 54588 different wines
- 28 columns (Wine, Country, Region, Subregion, Appellation, Vineyard, Grape, Critic_Score, Popularity, Price, Link, User_Score_Stars, User_Score_Status, User_Score_Ratings, User_Score_Expanded, Producer, Food, Style, ABV, Awards, Notes, Critic_Score_Expanded, Price_History, Price_Benchmark, Availability_History, Availability_Benchmark, Search_History, Search_Benchmark)