1. Predictive Analysis of Steel Fatigue Strength: a. Problem Statement: Addressed the challenge of predicting steel fatigue strength using machine learning. The primary issue tackled was multicollinearity in the dataset, crucial for accurate and reliable predictions in the mechanical domain.b. Approaches and Results: I. Addressed multicollinearity using PCA and Ridge Regression, achieving 94.44% testing accuracy.II. Conducted rigorous statistical analysis, including… Show more 1. Predictive Analysis of Steel Fatigue Strength: a. Problem Statement: Addressed the challenge of predicting steel fatigue strength using machine learning. The primary issue tackled was multicollinearity in the dataset, crucial for accurate and reliable predictions in the mechanical domain.b. Approaches and Results: I. Addressed multicollinearity using PCA and Ridge Regression, achieving 94.44% testing accuracy.II. Conducted rigorous statistical analysis, including Cook's Distance, Durbin-Watson test, and Variance Inflation Factor (VIF), to ensure the model's robustness.III. Recommended further optimization of Ridge Regression parameters, feature engineering, and exploration of ensemble methods for enhanced performancec. Ongoing Research: The work is part of an ongoing effort under DRDO, contributing to a research paper in the field2. Development of Property Prediction model for the hot rolled steel using Machine Learninga. Problem Statement: Developed algorithms to predict steel's mechanical properties (Ultimate Tensile Strength, Yield Strength, Percentage Elongation), bypassing time-intensive traditional tests, especially effective at high temperatures.b. Approaches and Results:I. Implemented Boruta Shap for precise feature selection, addressing the "curse of dimensionality" and optimizing model efficiency.II. Conducted meticulous model evaluation with Grid Search to determine the most effective hyper-parameter combinations.III. Identified Random Forest as the best model for Ultimate Tensile Strength and Yield Strength predictions, and XGBOOST for Percentage Elongation, outperforming traditional testing methods.c. Ongoing Research: Paper accepted by Machine Learning Research Journal, showcasing advancements in material science predictions. Show less