Recipe_Rating_Intel
- 0 Collaborators
This project aims to predict food ratings using machine learning models trained on various features extracted from recipes. By leveraging data such as ingredients, cooking methods, and user reviews, our models can provide accurate predictions of the ratings a recipe is likely to receive. ...learn more
Project status: Published/In Market
oneAPI, Artificial Intelligence
Intel Technologies
oneAPI
Overview / Usage
This project employs machine learning to predict recipe ratings, addressing challenges of subjectivity in evaluation and enabling recipe optimization. By analyzing ingredients, cooking methods, and user feedback, our models offer insights crucial for chefs, bloggers, and culinary platforms. In production, these models integrate seamlessly into recipe platforms, aiding in content curation, personalized recommendations, and recipe development. Ultimately, they enhance culinary experiences by providing standardized ratings and tailored suggestions, enriching user engagement and fostering a dynamic culinary community.
Methodology / Approach
Our methodology incorporates a diverse range of machine learning algorithms, including Logistic Regression, Decision Trees, Support Vector Machines (SVM), and XGBoost, each tailored to the task of predicting recipe ratings. These models are trained on a comprehensive dataset comprising recipe features and corresponding ratings, enabling them to learn patterns and relationships effectively. Leveraging industry-standard tools such as scikit-learn for logistic regression, decision trees, and SVM, as well as the XGBoost library for gradient boosting, we achieve notable accuracies ranging from 63% to 76%. Evaluation of model performance is conducted using the accuracy score metric, which measures the proportion of correctly predicted ratings. With an average accuracy score of 75%, our models demonstrate robust capability in accurately predicting food ratings, empowering users with valuable insights for culinary exploration and recipe optimization.
Technologies Used
The Recipe Rating project leverages essential elements such as machine learning methodologies for predictive modeling, supported by industry-standard libraries like scikit-learn and XGBoost. Additionally, the integration of Intel oneAPI Data Analytics Library (oneDAL) augments the project's capabilities with optimized algorithms and parallel processing capabilities. Python serves as the primary programming language, coupled with Jupyter Notebook for interactive development and experimentation. The Intel Distribution for Python ensures optimized performance on Intel hardware, while Git manages version control for collaborative development.