I began 2016 by compiling a list of popular articles from my blog in 2015. This "People's Choice" list contains many interesting articles, but some of my personal favorites did not make the list. Today I present the "Editor's Choice" list of articles that deserve a second look.
I've grouped the articles into four broad topics: SAS programming, computational statistics, statistical graphics, and matrix computations.
These two articles use Base SAS to accomplish some useful tasks:
- Many SAS procedures drop observations that contain a missing value for some variable. The remaining observations are called complete cases. If you are writing your own analysis, you need to know how to obtain complete cases in SAS.
- There are several ways to turn off ODS output in SAS. Learn five reasons you should use the ODS EXCLUDE statement to suppress SAS output.
For most analyses, you can call a built-in SAS procedure. However, I enjoy showing how to extend, modify, or explain a standard analyses by writing additional programming statements.
- Have you ever run a statistical procedure that does not converge to a solution? Learn about why some numerical optimization technique (such as Newton's method) diverge far from their initial states.
- Do you want to visualize the result of a multivariate regression? The regression coefficient plot displays the parameter estimates for a regression analysis.
- Do you use generalized linear models? Have you ever thought about the differences between a log-link model and a traditional OLS model? Read about how to visually compare two regression models that have different assumptions.
- Monte Carlo simulation is an important tool in the toolbox of the modern computational statistician. Learn how to perform Monte Carlo simulation in SAS for the analysis of contingency tables.
Although many SAS procedure automatically produce appropriate statistical graphics, it is sometimes necessary to use the SGPLOT or the Graph Template Language (GTL) to create your own graphics.
- Did you know that you can create a heat map by using PROC SGPLOT in SAS 9.4m3?
- For some data and some analyses, it is important to know how to preserve the aspect ratio of the data in a graph.
- For continuous data, it can be convenient to associate an ordinal scale such as "low-medium-high." Learn how to overlay a discrete color ramp on a histogram of a continuous variable.
From the many articles that I wrote about matrix computations, the following list presents three topics that SAS programmers often ask about on discussion forums:
- How to create an array of matrices in SAS.
- Does it make sense to perform matrix multiplication with missing values? If so, how do you multiply with missing values in SAS?
- The rank of a matrix is the number of linearly independent columns. What is the best way to compute the rank of a matrix in SAS?
There you have it, 12 topics that I think are worth a second look. Did I omit your favorite article from The DO Loop in 2015? Leave a comment.