Inspect Data
CauseMe does not store personal uploaded data. If you wish to contribute and share your data files, Zenodo.org allows you to upload, download and cite data in an easy way. Once uploaded to Zenodo, fill the following form so your data file can be openly accessed from CauseMe.
Success! Your data file has been shared
Check all the user uploaded data files here
lasar
Description
LASAR
(LASso
Auto-Regression).
Parameters
----------
data : ndarray
T (timepoints) x N (variables) input data
maxlags : int
Maximum number of lags to consider
n_subsamples : int
How often to subsample the data
subsample_sizes : ndarray
Possible sizes of the subsamples as fractions
cv : int
Number of cross-validation folds for the lasso variable selection step
aggregate_lags : function
Function that takes an N (to) x maxlags x N (from) ndarray as input and
outputs an N x N ndarray aggregating the lag-resolved scores,
for example
lambda x: x.max(axis=1).T
or
lambda x: x.sum(axis=1).T
Arguments for the common pre-processing steps of the data and the common
post-processing steps of the scores are documented in
utils.common_pre_post_processing
Returns
----------
scores : ndarray
Array where the (i,j)th entry corresponds to the link X_i --> X_j
Parameters
----------
data : ndarray
T (timepoints) x N (variables) input data
maxlags : int
Maximum number of lags to consider
n_subsamples : int
How often to subsample the data
subsample_sizes : ndarray
Possible sizes of the subsamples as fractions
cv : int
Number of cross-validation folds for the lasso variable selection step
aggregate_lags : function
Function that takes an N (to) x maxlags x N (from) ndarray as input and
outputs an N x N ndarray aggregating the lag-resolved scores,
for example
lambda x: x.max(axis=1).T
or
lambda x: x.sum(axis=1).T
Arguments for the common pre-processing steps of the data and the common
post-processing steps of the scores are documented in
utils.common_pre_post_processing
Returns
----------
scores : ndarray
Array where the (i,j)th entry corresponds to the link X_i --> X_j
Parameters
Adjust the parameters of the method
maxlags:
n_subsamples:
cv:
qrbs
Description
Perform
bootstrapped
ridge
regression
of
data
at
time
t
on
data
in
the
past
Parameters
----------
data : ndarray
T (timepoints) x N (variables) input data
lags : int
Number of lags to include in the modelling
alpha : double
Penalization parameter used for the ridge regression
q : double
The method performs 200 bootstrap samples, in each fitting a ridge
regression on a random subset of the data. This gives 200 estimates
of the effect i -> j.
We take the q'th quantile as the final estimate.
q = 1 corresponds to the max effect across samples, q = 0.5 to the
median effect.
n_resamples : int
Number of bootstrap samples drawn
Arguments for the common pre-processing steps of the data and the common
post-processing steps of the scores are documented in
utils.common_pre_post_processing
Returns
----------
scores : ndarray
Array with scores for each link i -> j
Parameters
----------
data : ndarray
T (timepoints) x N (variables) input data
lags : int
Number of lags to include in the modelling
alpha : double
Penalization parameter used for the ridge regression
q : double
The method performs 200 bootstrap samples, in each fitting a ridge
regression on a random subset of the data. This gives 200 estimates
of the effect i -> j.
We take the q'th quantile as the final estimate.
q = 1 corresponds to the max effect across samples, q = 0.5 to the
median effect.
n_resamples : int
Number of bootstrap samples drawn
Arguments for the common pre-processing steps of the data and the common
post-processing steps of the scores are documented in
utils.common_pre_post_processing
Returns
----------
scores : ndarray
Array with scores for each link i -> j
Parameters
Adjust the parameters of the method
lags:
alpha:
q:
n_resamples:
selvar
Description
SELVAR
(Selective
auto-regressive
model).
Parameters
----------
data : ndarray
T (timepoints) x N (variables) input data
maxlags : int
Maximum number of lags to include in the model.
If maxlags < 0 then the maximum lag will be iteratively
adjusted for each variable until no decrease in PRSS.
batchsize : int
Number of consecutive time points to use in each training batch.
If batchsize < 0 then batchsize is set to the maximum available
time boints given maxlags.
mxitr : int
Maximum number of iterations (each variable) for the
hill-climbing search. If mxitr < 0 then the
search will stop only when no decrease in PRSS is possible.
trace : int
If positive messages will be printed out during the search.
Arguments for the common pre-processing steps of the data and the common
post-processing steps of the scores are documented in
utils.common_pre_post_processing
Returns
----------
scores : ndarray
Array with scores for each link i -> j
pvalues : ndarray
Array with naively adjuste p-values of likelihood-ratio test,
should not be used for testing presence of edges
lags : ndarray
Array with estimated positive lags, 0 if no link
Parameters
----------
data : ndarray
T (timepoints) x N (variables) input data
maxlags : int
Maximum number of lags to include in the model.
If maxlags < 0 then the maximum lag will be iteratively
adjusted for each variable until no decrease in PRSS.
batchsize : int
Number of consecutive time points to use in each training batch.
If batchsize < 0 then batchsize is set to the maximum available
time boints given maxlags.
mxitr : int
Maximum number of iterations (each variable) for the
hill-climbing search. If mxitr < 0 then the
search will stop only when no decrease in PRSS is possible.
trace : int
If positive messages will be printed out during the search.
Arguments for the common pre-processing steps of the data and the common
post-processing steps of the scores are documented in
utils.common_pre_post_processing
Returns
----------
scores : ndarray
Array with scores for each link i -> j
pvalues : ndarray
Array with naively adjuste p-values of likelihood-ratio test,
should not be used for testing presence of edges
lags : ndarray
Array with estimated positive lags, 0 if no link
Parameters
Adjust the parameters of the method
maxlags:
batchsize:
mxitr:
trace:
slarac
Description
SLARAC
(Subsampled
Linear
Auto-Regression
Absolute
Coefficients).
Parameters
----------
data : ndarray
T (timepoints) x N (variables) input data
maxlags : int
Maximum number of lags to consider
n_subsamples : int
How often to subsample the data
subsample_sizes : ndarray
Possible sizes of the subsamples as fractions
missing_values : float
Values in the data matrix that equal missing_values are treated as
missing values
aggregate_lags : function
Function that takes an N (to) x maxlags x N (from) ndarray as input and
outputs an N x N ndarray aggregating the lag-resolved scores,
for example
lambda x: x.max(axis=1).T
or
lambda x: x.sum(axis=1).T
Arguments for the common pre-processing steps of the data and the common
post-processing steps of the scores are documented in
utils.common_pre_post_processing
Returns
----------
scores : ndarray
Array where the (i,j)th entry corresponds to the link X_i --> X_j
Parameters
----------
data : ndarray
T (timepoints) x N (variables) input data
maxlags : int
Maximum number of lags to consider
n_subsamples : int
How often to subsample the data
subsample_sizes : ndarray
Possible sizes of the subsamples as fractions
missing_values : float
Values in the data matrix that equal missing_values are treated as
missing values
aggregate_lags : function
Function that takes an N (to) x maxlags x N (from) ndarray as input and
outputs an N x N ndarray aggregating the lag-resolved scores,
for example
lambda x: x.max(axis=1).T
or
lambda x: x.sum(axis=1).T
Arguments for the common pre-processing steps of the data and the common
post-processing steps of the scores are documented in
utils.common_pre_post_processing
Returns
----------
scores : ndarray
Array where the (i,j)th entry corresponds to the link X_i --> X_j
Parameters
Adjust the parameters of the method
maxlags:
n_subsamples:
granger2d
Description
Perform
pairwise
Granger
causality
tests.
It uses test statistics as scores, pvalues are also returned.
Args:
data: matrix of observations (t x n)
maxlags: positive integer, the maximum lag considered in the VAR model.
test: name of the test used
code is adapted from https://github.com/cmu-phil/causal-learn
It uses test statistics as scores, pvalues are also returned.
Args:
data: matrix of observations (t x n)
maxlags: positive integer, the maximum lag considered in the VAR model.
test: name of the test used
code is adapted from https://github.com/cmu-phil/causal-learn
Parameters
Adjust the parameters of the method
maxlags:
test:
linear_var
Description
Fit
a
linear
VAR
model.
Return coefficients as edge scores,
additionally return (corrected) pvalues and lags.
Args:
data: matrix of observations (t x n)
maxlag: positive integer, the maximum lag considered in the VAR model.
correct_pvalues: if True
Return coefficients as edge scores,
additionally return (corrected) pvalues and lags.
Args:
data: matrix of observations (t x n)
maxlag: positive integer, the maximum lag considered in the VAR model.
correct_pvalues: if True
Parameters
Adjust the parameters of the method
maxlags:
correct_pvalues: