Converting array of Matplitlib dates to Datetime objects with NaN

Question

I have thousands of files with irregularly data that I am trying to interpolate to the same grid. The data is stored in individual numpy arrays for a number of different variables and times. I have been interpolating the times using matplotlib.dates by converting times to numbers. But because I don't want to extrapolate missing data the interpolation has introduced np.nan values in the dataset that makes the conversion from matplotlib date to datetime object tricky because I need to conserve shape (i.e. can't remove np.nans) but md.num2date cannot handle masked arrays or np.nan.

Is there another command like md.date2num and md.num2date in another library that can account for nans? Or is the only solution here a for loop?

import numpy as np
import matplotlib.dates as md

tdata = np.array([np.nan, np.nan, 736926.9827155,736926.98274578,736926.98276768,736926.98285067,736926.9828712])

#Allows num2date to work but removes the nan values and won't conserve shape
test1 = md.num2date(tdata[~np.isnan(tdata)])

print(len(tdata),len(test1))

#Using a mask will produce an error 
#test2 = md.num2date(np.ma.masked_array(tdata,np.isnan(tdata)))

Note that I need the shapes to be the same and that np.nan could occur anywhere in the array. I am using python 2.7

user7138814 · Accepted Answer · 2019-06-21 21:32:20Z

2

Maybe something like this?

test1 = np.full_like(tdata, np.nan, dtype=object)
test1[~np.isnan(tdata)] = md.num2date(tdata[~np.isnan(tdata)])

answered Jun 21, 2019 at 21:32

user7138814

2,05112 silver badges12 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Converting array of Matplitlib dates to Datetime objects with NaN

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related