Small error in Check Troughs function and naming convention #88
Description
Hi! Thank you for creating this tool - peakdet has been incredibly useful for my project so far!
I've noticed a couple of small things errors:
-
Very small thing is that when converting phys2neurokit (super useful function, thank you!), the data is saved with name f"{modality}_Raw" even if you have cleaned data if copy_data is set to True. This caused me some confusion - I just want others to know that the 'raw' is only in the name - if you have clean data, that's what gets copied!
-
utils/check_troughs function has a comparison sign instead of assignment which causes the manual peak correction to error, indicating that there is a mismatch between peaks and troughs and no index troughs[-1] exists. I've changed it like this and it works.
Original code:
def check_troughs(data, peaks, troughs=None):
"""
Confirms that troughs exists between every set of peaks in data
Parameters
----------
data : array-like
Input data for which `troughs` and `peaks` were detected
peaks : array-like
Indices of suspected peak locations in `data`
troughs : array-like or None, optional
Indices of suspected troughs locations in `data`, if any.
Returns
-------
troughs : np.ndarray
Indices of trough locations in `data`, dependent on `peaks`
"""
# If there's a trough after all peaks, keep it.
if troughs is not None and troughs[-1] > peaks[-1]:
all_troughs = np.zeros(peaks.size, dtype=int)
all_troughs[-1] == troughs[-1] ### ERROR HERE - could simply change to one =.
else:
all_troughs = np.zeros(peaks.size - 1, dtype=int)
for f in range(peaks.size - 1):
dp = data[peaks[f] : peaks[f + 1]]
idx = peaks[f] + np.argwhere(dp == dp.min())[0]
all_troughs[f] = idx
return all_troughs
How I've edited:
def check_troughs(data, peaks, troughs=None):
"""
Confirms that troughs exists between every set of peaks in data
Parameters
----------
data : array-like
Input data for which `troughs` and `peaks` were detected
peaks : array-like
Indices of suspected peak locations in `data`
troughs : array-like or None, optional
Indices of suspected troughs locations in `data`, if any.
Returns
-------
troughs : np.ndarray
Indices of trough locations in `data`, dependent on `peaks`
"""
# Handle edge cases
if peaks is None or len(peaks) == 0:
return np.array([])
if len(peaks) == 1:
return np.array([])
has_final_trough = (troughs is not None and
len(troughs) > 0 and
len(peaks) > 0 and
troughs[-1] > peaks[-1])
# If there's a trough after all peaks, keep it.
if has_final_trough:
all_troughs = np.zeros(peaks.size, dtype=int)
all_troughs[-1] = troughs[-1]
n_inter_troughs = peaks.size - 1
else:
all_troughs = np.zeros(peaks.size - 1, dtype=int)
n_inter_troughs = peaks.size - 1
for f in range(n_inter_troughs):
dp = data[peaks[f] : peaks[f + 1]]
min_idx = np.argmin(dp)
all_troughs[f] = peaks[f] + min_idx
# idx = peaks[f] + np.argwhere(dp == dp.min())[0]
# all_troughs[f] = idx
return all_troughs