Simulating Correlated Data with Different Types of Variables
- Burak AYDIN
- Mar 2, 2016
- 1 min read
This report reveals a small performance test for two data simulation methods. The utilized methods are expected to produce correlated continuous, categorical and binary variables based on a single multivariate normal distribution.
In order to test the performance, a fake dataset, let me name it kaf, was used. Please note that in a simulation study, a population covariance matrix, and descriptives of the variables (mean,sd and probabilities) will be enough.
Comments