DaO simulation

The DAG adaptation of the Onion method

Installation in Python

You can install daosim using pip:

pip install daosim

Example Usage in Python

import pandas as pd

from daosim import er_dag            # generates a Erdos-Renyi directed acyclic graph
from daosim import sf_out            # rewires a DAG to have scale-free out degree
from daosim import randomize_graph   # randomly shuffles the order of the variables
from daosim import corr              # samples a correlation matrix 
from daosim import simulate          # simulates a dataset
from daosim import standardize       # standardizes a dataset

p = 10    # number of variables
ad = 4    # average degree
n = 100   # number of samples

g = er_dag(p, ad=ad)
g = sf_out(g)
g = randomize_graph(g)

R, B, O = corr(g)
X = simulate(B, O, n)
X = standardize(X)

cols = [f"X{i + 1}" for i in range(p)]
df = pd.DataFrame(X, columns=cols)

Installation in R

To use the DaO in R include dao.R in your working directory.

Example Usage in R

p <- 10    # number of varibales
ad <- 4    # average degree
n <- 100   # number of samples

g <- er_dag(p, ad=ad)
g <- sf_out(g)
g <- randomize_graph(g)

model <- corr(g)

X <- simulate(model$B, model$O, n)
X <- standardize(X)