Simona’s presentation can be found here: http://www.r-gators.com/pdf/wildlife-tracking-data.pdf The RData file and RMarkdown file (used to make the rest of this post) can be downloaded at the GitHub repository here https://github.com/ufrmeetup/picardi. If you’re unfamiliar GitHub, you can download a zipped file by pressing the green “Clone or download” button on the right of that page. From there just unzip the folder and click on the .Rproj file (requires RStudio to be installed).
Check out Dr. Zhao’s presentation: http://www.r-gators.com/slides/zhao. This presentation was made using RMarkdown. The Markdown code is included below, however outputs from R (i.e. graphs) will only be displayed in the presentation. Download the RMarkdown file and an R script for generating data to test these scripts at the GitHub repository: https://github.com/ufrmeetup/zhao. Note the RMarkdown script relies on some private data, so it will not run on it’s own, but can be run using the simulated data.
In this article, we will be talking about the importance of cross-validation to choose the best model. To do this, we will show: the problem of using simple in-sample metrics for model selection with a univariate example the problem of using simple in-sample metrics for model selection with a multivariable example how to perform cross-validation using simple linear regression and some problems associated with this 1) The problem of using simple in-sample metrics for model seelction with a univariate example When dealing with a regression model, we are often interested in determining which covariates to keep in the model and which to through away.