Tutorial 6: Generating Transition Probability Matrices neurocaps.analysis.transition_matrix

The CAP.calculate_metrics function is designed to compute transition probabilities for all subjects and store these values into dataframes (one per group). These dataframes can subsequently be used as input for the transition_matrix function to generate visualizations of the average transition probabilities per group.

import numpy as np
from neurocaps.analysis import CAP, transition_matrix

# Simulate `subject_timeseries` structure
subject_timeseries = {str(x): {f"run-{y}": np.random.rand(10, 100) for y in range(1, 4)} for x in range(1, 11)}

cap_analysis.get_caps(
    subject_timeseries=subject_timeseries,
    cluster_selection_method="davies_bouldin",
    standardize=True,
    n_clusters=list(range(2, 6)),
)

2024-11-02 00:09:54,273 neurocaps.analysis.cap [INFO] [GROUP: All Subjects | METHOD: davies_bouldin] Optimal cluster size is 3.

outputs = cap_analysis.calculate_metrics(
    subject_timeseries=subject_timeseries,
    return_df=True,
    metrics=["transition_probability"],
    continuous_runs=True,
    output_dir=output_dir,
    prefix_filename="All_Subjects_CAPs_metrics",
)

print(outputs["transition_probability"]["All Subjects"])

Subject_ID	Group	Run	1.1	1.2	1.3	2.1	2.2	2.3	3.1	3.2	3.3
1	All Subjects	run-continuous	0.375	0.32142857142857145	0.30357142857142855	0.4074074074074074	0.38888888888888895	0.2037037037037037	0.3333333333333333	0.38461538461538464	0.28205128205128205
2	All Subjects	run-continuous	0.4363636363636364	0.3090909090909091	0.2545454545454545	0.28	0.43999999999999995	0.28	0.36363636363636365	0.2727272727272727	0.36363636363636365
3	All Subjects	run-continuous	0.3571428571428572	0.39285714285714285	0.25	0.39215686274509803	0.33333333333333326	0.27450980392156865	0.38095238095238093	0.2857142857142857	0.33333333333333337
4	All Subjects	run-continuous	0.41666666666666663	0.2	0.38333333333333336	0.46153846153846156	0.20512820512820507	0.3333333333333333	0.36	0.36	0.28
5	All Subjects	run-continuous	0.46153846153846156	0.2692307692307692	0.2692307692307692	0.26666666666666666	0.2666666666666666	0.4666666666666667	0.3269230769230769	0.34615384615384615	0.32692307692307687
6	All Subjects	run-continuous	0.33333333333333337	0.4444444444444444	0.2222222222222222	0.3103448275862069	0.3793103448275862	0.3103448275862069	0.2608695652173913	0.3695652173913043	0.36956521739130443
7	All Subjects	run-continuous	0.4067796610169492	0.3728813559322034	0.22033898305084745	0.39215686274509803	0.2941176470588236	0.3137254901960784	0.41025641025641024	0.3333333333333333	0.2564102564102564
8	All Subjects	run-continuous	0.33333333333333326	0.35185185185185186	0.3148148148148148	0.2708333333333333	0.33333333333333337	0.3958333333333333	0.46808510638297873	0.2765957446808511	0.25531914893617014
9	All Subjects	run-continuous	0.33333333333333337	0.3333333333333333	0.3333333333333333	0.43137254901960786	0.37254901960784315	0.19607843137254902	0.3409090909090909	0.29545454545454547	0.36363636363636365
10	All Subjects	run-continuous	0.3076923076923077	0.34615384615384615	0.34615384615384615	0.3793103448275862	0.4310344827586208	0.1896551724137931	0.358974358974359	0.38461538461538464	0.2564102564102564

kwargs = {"cmap": "viridis", "fmt": ".3f", "annot": True}

trans_outputs = transition_matrix(
    trans_dict=outputs["transition_probability"], show_figs=True, return_df=True, **kwargs
)

print(trans_outputs["All Subjects"])

From/To	CAP-1	CAP-2	CAP-3
CAP-1	0.3429448275257044	0.2881868830804245	0.368868289393871
CAP-2	0.3595098582449408	0.2914027330314906	0.3490874087235686
CAP-3	0.3259514626040708	0.3255982589778372	0.3484502784180919