ratal
commented
3 years ago Hi, Actually, interp() is used only for floating data. For integer or other types, the next data ('right') is taken. You can have a look at the method resample() in mdfreader.py, around line 702 For the moment, interp from numpy is used. However, I could add condition if interpolation kind is given to use interp1D from scipy instead. But I guess you want to force 'next option' even for floating ? I have a proposal for the interpolate function to be replaced by:
def interp_close_point(x, new_x, kind):
# interpolates with closest point at 'kind' side
idx = searchsorted(x, new_x, side=kind)
idx -= 1
idx = clip(idx, 0, idx[-1])
return idx
def interpolate(new_x, x, y, interpolation_kind):
# select right interpolation method depending of
# interpolation kind and data type
if interpolation_kind is None:
if y.dtype.kind == 'f':
return interp(new_x, x, y)
else:
return y[interp_close_point(x, new_x, 'right')]
else:
if interpolation_kind = 'previous':
return y[interp_close_point(x, new_x, 'left')]
elif interpolation_kind = 'next':
return y[interp_close_point(x, new_x, 'right')]
else
f = interp1d(x, y, kind=interpolation_kind)
return f(new_x)You could give a try
SanjayBalaji
Python version
3.7.4 (default, Aug 9 2019, 18:34:13) [MSC v.1915 64 bit (AMD64)]
Platform information
Windows-10-10.0.17763-SP0
Numpy version
1.18.2
mdfreader version
4.1
Description
Similar to that of interp1d in scipy, is there a way we specify the type of interpolation ('linear', 'quadratic', 'previous', 'next', etc..) being done during resampling of data? This is because, some channels record data only when there is a change( Eg. [ 7.00073242, 10.00061035, 19.99969482, 0.00061035, 19.99969482, 10.00061035, 19.99969482, 10.00061035, 19.99969482, 10.00061035, 10.00061035] ), but when we resampling, it interpolates linearly instead of interpolating in "next" method.
Thank you!