For goodness of fit's sake
March 15th, 2012 This week, we focus our attention to your burning questions about goodness-of-fit functions, like: What are the different functions for goodness of fit? How does each function differ from the rest? Which one do I use, and for what purpose?
Put simply, A goodness-of-fit function is a quantitative measure of the discrepancy (or the agreement) between the observed values and the values expected under a model in question. In general, a measure of goodness of fit helps us to find good (or optimal) values for a model's coefficients and facilitate the comparison of competing models in an effort to select the best one.
In this tutorial, we’ll start with the log-likelihood function (LLFi), and then expand to cover other derivative measures (e.g. Akaike's Information Criterion (AICi) and Bayesian/Schwarz Information Criterion (BICi/SICi/SBCi)). For an example, we use the time series of the ozone levels in downtown Los Angeles for the period between January 1955 and December 1972.
The exercise will leave you with a thorough understanding of goodness of fit, while demonstrating how NumXL can help you find an ideal model for your data.
For more details and a step-by-step tutorial, along with a downloadable spreadsheet and PDF, click this link: For goodness of fit's sake