Scribd
Upload
547 views6 pages

Variable Rationalization in Regression Models

The document discusses Variable Rationalization and Model Building in the context of Least Squares Estimation (LSE) for regression models. It outlines the steps for variable rationalization, key formulas for simple and multiple linear regression, model assumptions, and evaluation metrics. The focus is on ensuring relevant variables are selected and constructing valid statistical models for accurate predictions.

Uploaded by

basit
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
547 views6 pages

Variable Rationalization in Regression Models

The document discusses Variable Rationalization and Model Building in the context of Least Squares Estimation (LSE) for regression models. It outlines the steps for variable rationalization, key formulas for simple and multiple linear regression, model assumptions, and evaluation metrics. The focus is on ensuring relevant variables are selected and constructing valid statistical models for accurate predictions.

Uploaded by

basit
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd

Variable Rationalization & Model

Building in Least Squares Estimation


Introduction
• • Least Squares Estimation (LSE) is used to
estimate parameters in regression models.
• • Variable Rationalization ensures relevant,
independent variables are selected.
• • Model Building constructs a valid statistical
model with accurate predictions.
Variable Rationalization - Steps
• 1. Identifying Relevant Variables
• 2. Checking for Multicollinearity (Variance
Inflation Factor - VIF)
• 3. Handling Missing Data (Imputation,
Interpolation)
• 4. Scaling & Transformation (Normalization, Log
Transformation)
• 5. Feature Selection (Forward Selection, LASSO
Regression)
Least Squares Estimation Formulas
• Simple Linear Regression:
• • Slope (β₁) = Σ(xᵢ - x̄ )(yᵢ - ȳ) / Σ(xᵢ - x̄)²
• • Intercept (β₀) = ȳ - β₁x̄

• Multiple Linear Regression:


• • β = (XᵀX)⁻¹ XᵀY
Model Assumptions
• 1. Linearity (Relationship between X & Y must be
linear)
• 2. Independence (Errors should be uncorrelated)
• 3. Homoscedasticity (Constant variance of
residuals)
• 4. Normality of Errors (Residuals should be
normally distributed)
• 5. No Multicollinearity (Low correlation among
independent variables)
Model Evaluation Metrics
• • R²: Measures proportion of variance
explained by the model
• • Adjusted R²: Adjusts R² for the number of
predictors
• • Mean Squared Error (MSE): Average squared
error of predictions
• • AIC & BIC: Used for model comparison,
lower values indicate better models

You might also like