save a numpy array properly

Question

I am new to Python, so apologies if my question is very basic, but this is driving me crazy.

So, I created a 2D numpy array of floats, with a string row header. I would like to save the array so that I could re-open it and work it later.

I tried to use np.savetxt as follows:

np.savetxt('1_array', waves)

but it gives the obvious error:

TypeError: Mismatch between array dtype ('|S11') and format specifier ('%.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e %.18e')

I looked for a solution, but I don't seem to find any that would be appropriate for my case. Intuitively, there should be a way to specify the only the first row is a string, but I couldn't figure out how to do this.

Any help will be greatly appreciated!

I created a 2D numpy array of floats, with a string row header: it sounds like you're including both the floats and the strings into the same array (which results in everything being converted to strings). I'd suggest that you probably don't want to do that, and keep the header separate from the array (or use Pandas data frames). — NPE
– NPE, Commented Aug 8, 2017 at 11:59
Ok, but how could I keep the header separate from the array and save it at the same time? — RobertP.
– RobertP., Commented Aug 8, 2017 at 12:07
@DanielF I tried to do this: np.savetxt('1_array', waves, header=waves[0]), but it gives me the error: AttributeError: 'numpy.ndarray' object has no attribute 'replace' — RobertP.
– RobertP., Commented Aug 8, 2017 at 13:11
1. header needs to be a string, not a numpy array. 2. if waves[0] is your header, waves is still in string format, and you haven't split your data out. Also waves[0] is part of waves, so you're just duplicating your header. — Daniel F
– Daniel F, Commented Aug 8, 2017 at 13:17

Oluwafemi Sule · Accepted Answer · 2017-08-08 13:28:52Z

1

You need to save the header separate and save the rest as float.

waves_data = waves[1:].astype(float)
header = ' '.join(waves[0])

np.savetxt(
    '1_array', waves_data, delimiter=",", header=header
)

answered Aug 8, 2017 at 13:28

Oluwafemi Sule

39.3k1 gold badge63 silver badges88 bronze badges

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

Comments

Michał Machnicki · Accepted Answer · 2017-08-08 14:49:35Z

0

I don't understand why would you like to mix types in a single numpy array, but it's not really important here. Probably the simplest solution to save anything for later is using module pickle. Below piece of code which should solve you problem no matter what you put into the array:

import numpy as np
import pickle

waves = np.array([['a', 'b', 'c'], [1.0, 2.0, 3.0], [4.0, 5.0, 6.0]])  # if I understood correctly

# writing file to the hard drive
with open('testfile.p', 'wb') as f:
    pickle.dump(waves, f)

# reading from hard drive
with open('testfile.p', 'rb') as f:
    waves2 = pickle.load(f)

answered Aug 8, 2017 at 14:49

Michał Machnicki

3,0832 gold badges19 silver badges23 bronze badges

Collectives™ on Stack Overflow

save a numpy array properly

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related