def load_and_plot(df_files, write_dir='', plot_on=True):
'''Loads and plots the csv files in df_files. Optionally saves pdf.
Labels are used as the title'''
for i in df_files.index:
row = df_files.ix[i]
df = fop.load_emg(row.Path)
title = str(row.tolist())
fig, _ = viz.plot_emg(df, title=title)
if write_dir:
try: os.mkdir(write_dir)
except: pass
fig.savefig(os.path.join(write_dir, str(i) + '.pdf'))
if not plot_on:
viz.plt.close()
def quality(csv_path):
'''Calculate basic quality metrics for csv file.
* number of rows - to find files that are too short or long enough to sample multiple times
* max value across all channels (below 15,000) - magnitude of the signal
* median value across all channels - find files with signals that are too low
* number of spikes to 65535 - files where sensor was freaking out
* max fraction of zero values across channels - possibly dead sensors?
* max fraction of consecutive non-zero values - files with temporal binning issues
'''
df = fop.load_emg(csv_path)
len_df = len(df)
quality = {
"Length": len_df,
"Max": df[df < 15000].unstack().max(),
"Median": df.unstack().median(),
"N_spikes": (df > 65000).sum().sum(),
"MaxFrac_zero": max((df == 0).sum().divide(len_df) * 100),
# "MaxFrac_repeat": max(((df != 0) & (df.diff() == 0)).sum().divide(len_df) * 100)
"MaxFrac_repeat": max(((df > 100) & (df.diff() == 0)).sum().divide(len_df) * 100)
}
return quality