Automated Penalized Regression Pipeline (mlr3)

An institutional-grade machine learning pipeline built using the R mlr3 ecosystem and glmnet. This project is engineered to handle high-dimensional datasets by deploying regularized linear models, specifically focusing on the optimization and deployment of Lasso Regression.

---

Key Features

• Lasso Regression Supremacy: Implements pure L1 penalty constraints (alpha = 1) to execute automatic feature selection. Lasso enforces sparsity by shrinking non-essential coefficients to absolute zero, proving to be the optimal statistical model for variance reduction.
• MSE Validation Architecture: Calibrates and validates model predictive superiority strictly against the Mean Squared Error (MSE) metric to guarantee minimal prediction residual variance.
• Modern R ML Ecosystem: Utilizes the object-oriented mlr3 framework for strict data partitioning, task instantiation, and predictive modeling workflows.
• SQL Compliance Logging: Integrates the RSQLite and DBI libraries directly into the R pipeline to capture and persist historical model performance metrics (MSE, RMSE, retained features) into an embedded relational database.

---

Technical Stack & Dependencies

• Language: R
• Machine Learning Framework: mlr3, mlr3learners, mlr3pipelines
• Regularization Engine: glmnet
• Database Architecture: DBI, RSQLite

---

Local Execution Guide

1. Clone the Repository:

   git clone [https://github.com/dimssrmdn01/regularized-regression-mlr3.git](https://github.com/dimssrmdn01/regularized-regression-mlr3.git)
   cd regularized-regression-mlr3

2. Install R Dependencies: Execute the following inside your R console:

   install.packages(c("mlr3", "mlr3learners", "mlr3pipelines", "glmnet", "DBI", "RSQLite"))

3. Run the Analytical Pipeline: Execute the core script via the terminal:

   Rscript regularized_pipeline.R