ddn3.tools
Utility functions of DDN
Module Contents
Functions
|
Standadize each column of the input data |
|
Generate zero mean multivariate data and standardize it. |
|
Several method to concatenate two arrays |
|
Threshold the input matrxi to get adjacency matrix |
|
Calcualte common and differential network from the output of DDN |
|
Objective function for DDN, assuming two conditions having the same sample size |
- ddn3.tools.standardize_data(data, scaler='std', zero_mean_already=False)
Standadize each column of the input data
- Parameters:
data (ndarray) – Input data. Each column is a feature.
scaler (str, optional) – Method to calculate the scale of each feature, by default “std”
zero_mean_already (bool, optional) – The data is already with zero mean, by default False
- Returns:
standard_data – Standardized data
- Return type:
ndarray
- ddn3.tools.gen_mv(cov_mat, n_sample)
Generate zero mean multivariate data and standardize it.
- Parameters:
cov_mat (ndarray) – Covariance matrix
n_sample (int) – Sample size
- Returns:
Generated samples
- Return type:
ndarray
- ddn3.tools.concatenate_data(controldata, casedata, method='diag')
Several method to concatenate two arrays
- ddn3.tools.get_net_topo_from_mat(mat_prec, thr=0.0001)
Threshold the input matrxi to get adjacency matrix
The input is also made symmetry. As we do not use self loop, the diagonal elements are always 0.
- Parameters:
mat_prec (ndarray) – Input matrix. It could be a precision matrix, or a coefficient matrix.
thr (float, optional) – Threshold to remove too small items in mat_prec, by default 1e-4
- Returns:
Adjacency matrix
- Return type:
ndarray
- ddn3.tools.get_common_diff_net_topo(g_beta, thr=0.0001)
Calcualte common and differential network from the output of DDN
g_beta[0] is the coefficient matrxi for condition 1 g_beta[0] is the coefficient matrxi for condition 2
- ddn3.tools.ddn_obj_fun(y, X, lambda1, lambda2, n1, n2, beta)
Objective function for DDN, assuming two conditions having the same sample size