OVINO is an end-to-end project built, evaluated, and deployed by me, Stephen. My app was competitively selected as best data science project by The Data Incubator (TDI) faculty amongst two cohorts of doctorates (about 50 fellows).
In short, OVINO helps me find well-priced wines at my nearest stores by crowdsourcing user reviews containing prices. It also provides personal wine recommendations based on a Vivino user's history of wine ratings using a wide & deep recommender.
The motivations and designs for OVINO are described in this About section.
Code designs and data analyses are available in detail here:
- Web scraping and databasing the LCBO: Repo
- Exploratory Data Analysis of 31 million wine reviews: Blog post & code
- Data analysis of my price-sentiment training set: Blog post & code
- NLP (AI language model) and Machine Learning model trainings (LogReg, Naive Bayes, SVM, RF, XGBoost, LightGBM): Blog post & code
- Wide & Deep Neural Net Recommender for personalized wines: Implemented in OVINO! Blog post in progress.
Distributed under the BSD 3-Clause License. See LICENSE.txt for more information.
Stephen Ro