Data Classification(KNN, Bayes Classifier with unimodal Gaussian Density)
Repository Title: Steel Plate Fault Classification with KNN and Bayesian Methods
Repository Description: Welcome to the Steel Plate Fault Classification repository! This project showcases the application of machine learning techniques to classify steel plate faults into two distinct categories. The analysis employs K-Nearest Neighbors (KNN) and Bayesian classification algorithms, providing valuable insights into their performance on the given dataset.
Highlights:
Data Exploration and Preprocessing: Dive into the world of steel plate fault classification by loading the dataset, splitting it into training and testing sets, and saving the processed data for future use.
K-Nearest Neighbors (KNN) Classification: Uncover the power of KNN as we employ it with different 'k' values (1, 3, and 5). Witness the creation of confusion matrices and the computation of classification accuracy for each iteration.
Data Normalization: Gain a deeper understanding of the importance of data normalization. Experience the transformation of features using min-max scaling and observe the impact on KNN classification accuracy.
Bayesian Classifier: Explore the realm of probabilistic modeling with the Bayesian classifier. Understand how class-wise means, covariances, and prior probabilities contribute to accurate predictions. Witness the elegance of Bayesian decision-making through confusion matrices and accuracy scores.
Result Comparison: Unearth the ultimate showdown! Witness the battle of the algorithms as we present a comprehensive table showcasing the accuracy achieved by KNN (k=5), normalized KNN (k=3), and the Bayesian classifier.