In [ ]:

    

import classify_covs
import show_connectomes
import covariance
import matplotlib.pyplot as plt
from sklearn.covariance import EmpiricalCovariance, LedoitWolf
import itertools
#%matplotlib inline


    

In [ ]:

    

def plot_results(t_df, p_th=.05, estim_title=None):
    # p-val at which to threshold
    p_th = .05
    if estim_title is not None:
          estim_title = " ({})".format(estim_title)
    for ix_ in range(len(t_df)):
        tstats = covariance.vec_to_sym(t_df["tstat"].iloc[ix_])
        pvals = covariance.vec_to_sym(t_df["pval"].iloc[ix_])
        pvals = correct(pvals, fdr) 
        tstats[pvals > p_th] = 0.
        title = t_df["comparison"].iloc[ix_] + estim_title
        show_connectomes.plot_adjacency(tstats, n_clusters=1,
                                        title=title,
                                        vmin=None, vmax=None, col_map="red_blue_r",
                                        save_fig="/home/storage/workspace/parietal_retreat/covariance_learn/figures/" +
                                        title + ".pdf")


    

In [ ]:

    

base_estimators = [EmpiricalCovariance(assume_centered=True), LedoitWolf(assume_centered=True)]
estimators_ = list(itertools.product(["tangent", "partial correlation", "correlation"], base_estimators))
est = ('kind', 'base_estimator')
estimators = [dict(zip(*[est, e])) for e in estimators_]
t_test = list()
for est_ in estimators:
    t_test.append(classify_covs.statistical_test(root_dir="/home", estimators=est_, verbose=0))
    plot_results(t_test[-1], estim_title=est_["kind"] + " " + format(est_["base_estimator"]).split("(")[0])