I am using sklearn’s non-negative matrix factorisation (NMF) to decompose my fmri data into time series (a.k.a latent features(k); size: t x k; t: time points) and spatial maps (size: k x N; N: number of ROIs) and I would like to construct a 3D surface brain map using the spatial maps using different numbers of latent features (k).
Here is my code:
import pandas as pd
from sklearn.decomposition import NMF
import scipy.stats as stats
model = NMF(n_components=5,init = "nndsvda")
W = model.fit_transform(fMRI_data)
H = model.components_
I am not able to plot the NMF values (here H) on to the brain surface. Maybe I am missing something here