This repository contains all the best submission filed from Practise Hackathon conducted by iitg.ai club for their Machine Learning course.
I started my approach by exploring the provided training and testing datasets along with the sample submission. The goal was to predict a target variable based on anonymized features. I performed data preprocessing steps to handle missing values, encode categorical variables, and scale numerical features using techniques like StandardScaler.
To improve the prediction accuracy, I explored ensemble learning techniques. I used stacking to create ensemble models that combined SVR, XGBoost, Linear Regression, and AdaBoost. By leveraging the strengths of different models, I aimed to improve the overall performance. I experimented with various models including SVR, Linear Regression, XGBoost, AdaBoost, and Random Forest. I trained these models on the preprocessed training data and evaluated their performance using metrics like RMSE on the validation set. After evaluating the performance of the models on the validation set using metrics like RMSE, I selected the best-performing model or ensemble for generating predictions.
Throughout the hackathon, I focused on iteratively experimenting with different models, preprocessing techniques, and ensemble learning strategies. The aim was to find the best combination that minimized the RMSE and improved prediction accuracy. By employing various techniques and methodologies, I aimed to leverage the strengths of different models and optimize the overall performance of my solution.