Regression - AIC/SBC Comparison

In summary, the conversation is about comparing AIC/SBC values between different models. The speaker ran a linear regression and a log-linear regression, with the log-linear regression producing more negative values. The textbook suggests that smaller values are better, but only compares positive values that tend towards zero. The question is whether the absolute values should be compared or if a strictly greater than scheme should be used. The speaker later figures out a solution and no longer needs help.
  • #1
TheBestMilk
13
0
I'm not sure if this is the right place for this question, but it was on the comparison between different model's AIC/SBC values.

I ran a linear regression and got an AIC/SBC of .743/.768. When I ran the same regression in log-linear form I ended up with an AIC/SBC of -7.559/-7.534.

My textbook suggests that the smaller the value, the better the model, but it only compares positive values which seem to tend towards zero.

My question is this: should I be comparing the absolute values of these (so that the closet to zero is the best) or should I be looking at a strictly greater than scheme in which the more negative, the smaller, and therefore the better?

Thanks!
 
Physics news on Phys.org
  • #2
Never mind. I figured out I could estimate the log-linear model in levels which would allow me a comparison that fits with the generic linear regression. I couldn't figure out how to delete the post. Thanks anyways!
 

FAQ: Regression - AIC/SBC Comparison

1. What is AIC and SBC in the context of regression?

AIC (Akaike Information Criterion) and SBC (Schwarz-Bayesian Criterion) are both statistical measures used to evaluate the goodness of fit of a regression model. They take into account both the model's complexity and its predictive power.

2. How do AIC and SBC differ from each other?

While both AIC and SBC aim to find a balance between model complexity and predictive power, they use different approaches. AIC penalizes models based on their complexity, while SBC takes into account the number of parameters in the model. As a result, AIC tends to favor simpler models, while SBC may favor more complex models.

3. Which criterion should I use for model selection, AIC or SBC?

There is no definitive answer to this question as it depends on the specific data and research question. Some researchers argue that AIC is more appropriate for predictive modeling, while SBC is better for explanatory modeling. It is recommended to compare the results from both criteria and consider the context of the study before making a decision.

4. What is the interpretation of AIC and SBC values?

The lower the AIC or SBC value, the better the model fits the data. A difference of 2 or more between two models' AIC or SBC values indicates that the model with the lower value is significantly better at explaining the data.

5. Can AIC and SBC be used for non-linear regression models?

Yes, AIC and SBC can be used for any type of regression model, including non-linear models. However, they may not perform as well as other criteria specifically designed for non-linear models, such as the Bayesian Information Criterion (BIC). It is important to consider the assumptions and limitations of these criteria when applying them to non-linear models.

Similar threads

I Linear regression, feature scaling, and regression coefficients
Replies
8
Views
2K
A What is the reasoning for this prediction?
Replies
4
Views
1K
I Linear regression and random variables
Replies
30
Views
3K
I How to propagate errors through a regression & non-linear model?
Replies
8
Views
2K
B Comparing Approaches: Linear Regression of Y on X vs X on Y
2
Replies
64
Views
4K
Linear Least-Squares Regression: a_0 = ?
Replies
3
Views
2K
I Regression line with zero slope and average as best prediction
Replies
2
Views
1K
How Do You Calculate Predicted Values in Least Squares Regression?
Replies
10
Views
2K
Proving the convergence of series
Replies
14
Views
2K
Linear and Non-Linear Regression
Replies
1
Views
1K

Hot Threads

Recent Insights

Back
Top