Model Validation & Diagnosis Main §

---

Main: §

• Inference on the regression parameters
• Examine through plots
• Testing
• Multicollinearity
• Diagnose for out lier detcetion

---

Diagnosis : Detection:

• Outliers
• Multicollinearity

For predictor X:

1. Dot plot
2. Stem and Leaft plot
3. Box Plot
4. Sequence Plot(if the observations are time ordered

Diagnostics for Residulas(): Residulas

Violation of Assumptions:

1. Regression funciton is non-Linear
2. Non-cosntant variance
3. Error terms not independent
4. Possibility of Outliers
5. None Normal Distribution of errors
6. Omissiion of some of the important predictors

---

Key : §

1. Avoid the variance of the response value is a linear function
   1. we would not know if the relation is from mean or vairance(we should see varaince as a constant for the varaince of the resposne value)
   2. it’s a free for all problem

---

TAGS §