F-statistic of manner separability¶

import neuraspeech as ns
for k, v in ns.PHON2MANNER.items():
    print(f'{k}: {v}')

# Load in the PRP data
prps = ns.PRPData.from_pickle('examples\saved\prps_saved.pkl')

# Compute F
fstats = ns.FStatistic(
    prps,                   # Pass the PRPData object here
    by = 'filename',        # Default to 'filename', meaning that the results will be computed separately for each filename value
    category = 'manner'     # Defualt to 'manner'. Name of variable indicatings cateogry labels.
    )

# Save the F-stat results
fstats.save('examples/saved/fstats_saved.pkl')

# Load it again
fstats = ns.FStatistic.from_pickle('examples/saved/fstats_saved.pkl')

fstats_data = fstats.get_data()
print(fstats.get_data())

print(fstats_data['F'])

# Recode levels "pres" and "int" using a dictionary
recode_pres = {'bi': 'Binaural', 
               'mo': 'Monaural'}

fstats.recode('pres', recode_pres)

recode_int= {'60': '60_dB',
             '75': '75_dB'}

fstats.recode('int', recode_int)

print(fstats.get_data().head())

# Fronto-central channels
fronto_central_chs = ['AF3', 'AF4', 'Fz', 'F3', 'F4', 'FC1', 'FC2']

# Mark these time points and their corresponding text labels
mark = {'time': [0.05, 0.12, 0.23, 0.4], 'label': ['R1', 'R2', 'R3', 'R4']}

# Plot the averaged PRP for each manner, divided by the factors in split_by
fstats.plot_f_stats(
    split_by = ['pres', 'int'],                  # Factor(s) used to divide the figure.
    title_prefix = '(FC)',                       # Optional, append a prefix to the title of each subplot
    subset_sensors = fronto_central_chs,         # Plot the average F-stat in this subset of electrodes
    equal_y_scale = True,                        # Ensure that the y-scale is the same across subplots
    se = 1,                                      # SE around mean
    lw = 3.5,                                    # Linewidth
    mark = mark,                                 # Mark these time points
    show_topo = True,                            # Show topo of the plotted sensors
    line_colors = '#008ADC',                   # Line color, default is red,
    fig_path = 'examples/figures/fstats_fig.svg' # Optional. Save the figure to disk by giving its path
    )

fstats.plot_f_topo(
    split_by = ['pres', 'int'],                 # Split the figure based on combinations of the levels of these factors
    times = [0.05, 0.12, 0.23],                 # Plot topomaps of F at these time points
    vlim = (0.0, 6.0),                          # Lower and upper bounds of the colormap
    cmap = 'Reds',                              # Matplitlib colormap. Default is "Reds"     
    fig_path = 'examples/figures/f_topo.svg'
    )

fstats_df = fstats.get_values(
    times = [(0.04, 0.06), (0.10, 0.12), (0.19, 0.21), (0.39, 0.41)], 
    subset_sensors = fronto_central_chs,
    summarize_sensors = True,
    summarize_time = True, 
    summary_func = 'mean',
    output_col_labels = ['R1', 'R2', 'R3', 'R4']
    )
fstats_df.head(3)

import eelbrain
fstats_data = fstats.get_data()
fstats_data['F'].sensor.set_adjacency(connect_dist = 1.6) # Setting adjacency defining which electrodes are neighbors
fstats_data['sub'].random = True                          # Subject as a random effect

# ANOVA: F ~ ses * int * pres
fstats_anova_res = eelbrain.testnd.ANOVA(
    'F',
    'ses * int * pres',
    data = fstats_data,
    pmin = 0.05,        # Use uncorrected p = 0.05 as threshold for forming clusters
    tstart = 0.0,
    tstop = 0.5,
    samples = 10000     # Number of permutations to speed up the example; use 10'000 when the exact p-value matters
)
fstats_anova_res.find_clusters(0.05)