description of code and tools to compute higher order psychological networks
- Input: 2D data (observations x variables). Real data (here) you find plenty, or simulations (see below)
- Clone this repo and run
this function, with the command
[Otot, O_val_size_tot] = hoi_exhaustive_loop_zerolag_all(data, max_size, biascorrection);, where
max_size = highest multiplet order (e.g. 4)
biascorrection = whether the entropy is corrected for sampling bias (1 = true)
to statistically validate all the multiplets, or
this function, with the command
[Otot, O_val_size_tot] = hoi_exhaustive_loop_zerolag(data, max_size, n_highest, biascorrection);, where
max_size = highest multiplet order (e.g. 4)
n_highest = number of multiplets with highest informational value that is retained for boostrap validation (e.g. 50)
biascorrection = whether the entropy is corrected for sampling bias (1 = true)
to statistically validate only a certain number with highest informational value.
For hardcore python users, this repo is a functioning, work in progress version, but slower and still without the important check of carryover significance
3. Plot the output using this repo and the notebook High order network psychometrics.ipynb
Use the R files to reproduce data with structures according to the figure below
