# Data organization¶

Skbold is quite flexible in how to work with data organized in different ways. However, to optimally use the Mvp classes (especially the MvpBetween class), it’s best to organize your data hierarchically. For example, suppose you have a .feat-directory with first-level pattern estimates of two two runs (‘run-1’, ‘run-2’) for 3 subjects (‘sub-01’, ‘sub-02’, ‘sub-03’). A nice organization would be as follows:

project_dir
├── sub-01
│   ├── run-1
│   └── run-2