Cofee Quality Prediction Using Machine Learning

The analysis and modeling of a dataset related to sensory evaluations of coffee quality scores is presented in this report. The report explores various features related to aroma, flavor, aftertaste, acidity, body, balance, uniformity, cleanliness, sweetness, and overall quality of coffee. The report starts with data preprocessing, including handling missing values and encoding categorical variables. The report then drops irrelevant columns and extracts useful information from date columns. For the modeling phase, the report splits the data into training and testing sets and applies feature scaling using the MinMaxScaler to normalize the numeric features. The report then trains a Linear Regression model and evaluates its performance using mean squared error (MAE) and R- squared score. The model achieved a low MAE of 0.2567 and a high R-squared score of 0. 9476, indicating a good fit to the data. Furthermore, the report compares the performance of various regression models using cross-validation. The models included Linear Regression, Ridge, Lasso, ElasticNet, Decision Tree Regressor, Random Forest Regressor, Gradient Boosting Regressor, Support Vector Regressor (SVR), and MLP Regressor. Finally, the report recommends the best coffee combination based on the analysis. The recommended coffee combination is from Coffee Quality Union, with the variety of Castillo, the country of origin of Colombia, the farm name of Finca El Paraiso, and the processing method of Double Anaerobic Washed.