Regression - AIC/SBC Comparison

Click For Summary
SUMMARY

The discussion centers on the comparison of Akaike Information Criterion (AIC) and Schwarz Bayesian Criterion (SBC) values derived from linear and log-linear regression models. The user initially reported AIC/SBC values of 0.743/0.768 for a linear regression and -7.559/-7.534 for a log-linear regression. The conclusion reached is that when comparing these values, one should focus on the absolute values, with smaller (more negative) values indicating a better model fit. The user ultimately resolved their query by estimating the log-linear model in levels for a more straightforward comparison.

PREREQUISITES
  • Understanding of linear regression analysis
  • Familiarity with Akaike Information Criterion (AIC) and Schwarz Bayesian Criterion (SBC)
  • Knowledge of log-linear modeling techniques
  • Ability to interpret regression output and model fit statistics
NEXT STEPS
  • Research the implications of AIC and SBC in model selection
  • Learn how to estimate log-linear models in statistical software
  • Explore the differences between linear and log-linear regression models
  • Study the interpretation of regression diagnostics and model fit metrics
USEFUL FOR

Statisticians, data analysts, and researchers involved in model selection and regression analysis will benefit from this discussion, particularly those working with AIC and SBC for evaluating model performance.

TheBestMilk
Messages
12
Reaction score
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
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!
 

Similar threads

Undergrad Linear regression, feature scaling, and regression coefficients
  • · Replies 8 ·
Replies
8
Views
3K
Graduate What is the reasoning for this prediction?
  • · Replies 4 ·
Replies
4
Views
1K
Logarithmic Regression Question (TI-84 Plus Graphing Calculator)
  • · Replies 4 ·
Replies
4
Views
2K
Undergrad Linear regression and random variables
  • · Replies 30 ·
2
Replies
30
Views
4K
Undergrad How to propagate errors through a regression & non-linear model?
  • · Replies 8 ·
Replies
8
Views
3K
High School Comparing Approaches: Linear Regression of Y on X vs X on Y
  • · Replies 64 ·
3
Replies
64
Views
5K
Linear Least-Squares Regression: a_0 = ?
  • · Replies 3 ·
Replies
3
Views
3K
Undergrad Regression line with zero slope and average as best prediction
  • · Replies 2 ·
Replies
2
Views
2K
How Do You Calculate Predicted Values in Least Squares Regression?
  • · Replies 10 ·
Replies
10
Views
2K
Proving the convergence of series
  • · Replies 14 ·
Replies
14
Views
2K