91Ó°ÊÓ

The Coefficient of Determination, often represented as \( R^2 \), is a key metric in regression analysis. It indicates how well the independent variables explain the variability in the dependent variable. In simple terms, it tells us how good our model is at making predictions.

### Understanding \( R^2 \)
The formula to calculate \( R^2 \) is:

• \[ R^2 = \frac{\text{SSR}}{\text{SST}} \]

Here, \( \text{SSR} \) (Sum of Squares Regression) represents the portion of total variability explained by the model, while \( \text{SST} \) (Sum of Squares Total) is the total variability in the data.

For example, in our data where \( \text{SST} = 15,182.9 \) and \( \text{SSR} = 14,052.2 \), the \( R^2 \) value is approximately 0.925. This means that 92.5% of the variability in our dependent variable, \( y \), is explained by the model. Therefore, this provides a good indication of a strong model fit. The higher the \( R^2 \), the better the model explains the data variability.

While the Coefficient of Determination \( R^2 \) is useful, it has some limitations, especially when multiple predictors are involved. This is where Adjusted R-squared (\( R^2_a \)) comes into play.

### Why Adjusted R-squared?
This adjusted metric provides a more accurate measure by considering the number of predictors relative to the number of observations. Including more variables in the model generally inflates \( R^2 \), but \( R^2_a \) corrects this by also considering the model complexity.

### Calculating \( R^2_a \)
The formula to compute \( R^2_a \) is:

• \[ R^2_a = 1 - \left(1 - R^2\right) \frac{n - 1}{n - p - 1} \]

where \( n \) is the total number of observations, and \( p \) is the number of predictors.

In our exercise, with \( n = 10 \) and \( p = 2 \), the Adjusted R-squared is approximately 0.905. Though slightly lower than \( R^2 \), it offers a more realistic view by penalizing the model complexity, ensuring that adding noise variables doesn’t make the model look deceptively strong.

Sum of Squares is integral to understanding variability in data within regression analysis.

### Types of Sum of Squares
1. **Total Sum of Squares (\( \text{SST} \))**: This measures the total variance in the data. It's the base level from which all other sums of squares distill their portion.
- \( \text{SST} = 15,182.9 \) in our example. 2. **Regression Sum of Squares (\( \text{SSR} \))**: This indicates the variance explained by the model itself.
- Here, \( \text{SSR} = 14,052.2 \), meaning a large part of the total variance is explained by the regression model.
### Importance in Analysis
By using \( \text{SST} \) and \( \text{SSR} \), we determine how well the model fits via \( R^2 \) and develop a deeper insight into the model’s predictive power. Each part of the Sum of Squares calculation lets us measure and compare how much worse or better our model could perform.