A Comparative Analysis of Machine Learning Models for Classifying Sanora, Mamra, and Regular Almond Varieties

This work categorizes almonds into three classes, namely, SANORA, MAMRA, and REGULAR, with the help of a different machine learning model. Several features of almonds are included in this dataset: length, width, thickness, area, perimeter, roundness, solidity, compactness, aspect ratio, eccentricity, extent, convex hull derived from image data. Applied these models on the given dataset for the performance anticipation of distinct models. Logistic regression was 65.48 %, K-Nearest Neighbors reached 78.65%, Random Forest performed at 86.83 %, XGBoost reached 83.27 %, and LightGBM was <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{8 5. 4 1 \%}$</tex>. After some further optimizations using the best parameters, Random Forest attained the best accuracy among the models at <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{8 7. 1 9 \%}$</tex>. This classification work includes the three most important classes. The metric scores with optimized parameters of Random Forest are as follows: for class 0, the precision is 0.93 and recall is 0.92; class 1, the precision is 0.83 and recall is 0.82; class 2, the precision is 0.84 and recall is 0.86. It gave an overall accuracy of 87 %. The dataset used was from Kaggle, and it had a rich set of features derived from images, hence suitable for supervised classification techniques. Among these, the Random Forest model, optimized using grid search, was supreme among them in terms of classifying almonds into the target categories. This present work pendicles on the efficiency of some advanced machine learning algorithms useful in agricultural product classification.