The Paper “Potato Leaf Disease Detection using Machine Learning” focuses on the accurate identification of diseases in potato leaves to support early intervention and enhance crop yield. The process begins with Data Collection, where a dataset of 2,170 potato leaf images is sourced from Kaggle. This dataset is categorized into three classes: Early Blight, Late Blight, and Healthy. Next, Data Pre-Processing is undertaken to ensure the images are cleaned, resized, and normalized, preparing them for effective analysis. During Feature Extraction, relevant features from the images are identified to represent the data meaningfully. The dataset is then split into two subsets: the Train Image Set with 1,736 images, and the Test Image Set with 434 images, to facilitate model training and evaluation respectively. Various Machine Learning Techni-ques are applied, including k-Nearest Neighbors (KNN), Support Vector Machine (SVM), Naive Bayes, Decision Tree, and Random Forest. Each trained model undergoes rigorous testing to evaluate its performance. To assess the effectiveness of each model, Performance Metrics such as accuracy, precision, recall, and F1-score are computed. The model exhibiting the highest accuracy is subjected to Feature Selection using Neighborhood Component Analysis (NCA) to enhance its performance further. Ultimately, a Hybrid Model is developed by combining the strengths of the individual models, aiming to improve overall accuracy and robustness in disease detection. This comprehensive approach integrates multiple machine learning techniques and feature selection methods, offering a robust solution for potato leaf disease detection. The comproject demonstrates the potential of machine learning in agricultural applications, contributing to more efficient and precise disease management.