convert elements of an array from scientific notation to decimal notation in python

Question

I have a numpy array come of whose elements are in scientific format and I want to convert them into decimal format. My numpy array looks like this:

[array([ 93495052.96955582,  98555123.06146193])]
[array([  1.00097681e+09,   9.98276347e+08])]
[array([  6.86812785e+09,   6.90391125e+09])]
[array([  7.75127468e+08,   8.02369833e+08])]

and this is formed using this line in my code:

list1.append(np.array(regr.predict(data),dtype = np.float))

Now I want to convert elements in list1 from scientific format to decimal format. I looked around for some solution and found out that print format(0.00001357, 'f') converts numbers from scientific format to decimal format but how do I use it to convert elements of my array?

Answer 1

First off, as several people have noted, there's a very large difference between how the numbers are displayed and how they're stored.

If you want to convert them to strings, then use '{:f}'.format(x) (or the % equivalent).

However, it sounds like you're only wanting the numbers to be displayed differently when you're working interactively (or through a print statement).

Changing how numpy arrays are printed

The way that numpy arrays are displayed interactively is controlled by numpy.set_printoptions.

Note that this does not convert the numbers to strings or change them in any way.

As a quick example:

In [1]: import numpy as np

In [2]: x = 1e9 * np.random.random(5)

In [3]: x
Out[3]:
array([  4.96602724e+08,   5.42486095e+08,   4.74495681e+08,
         7.37709684e+07,   9.75410927e+08])

In [4]: np.set_printoptions(formatter={'float_kind':'{:f}'.format})

In [5]: x
Out[5]:
array([496602723.824146, 542486095.316912, 474495680.688025,
       73770968.413642, 975410926.873148])

We've only changed how numpy will display the numbers. They're still floats.

We can operate on them mathematically, and they'll behave like numbers:

In [6]: x[0]
Out[6]: 496602723.82414573

In [7]: x[0] * 2
Out[7]: 993205447.64829147

Converting to strings

Now let's say we had converted them to a list of strings:

In [1]: import numpy as np

In [2]: x = 1e9 * np.random.random(5)

In [3]: x
Out[3]:
array([  2.56619581e+08,   2.55721261e+08,   3.36984986e+08,
         2.67541556e+08,   9.01048842e+08])

In [4]: x = ['{:f}'.format(item) for item in x]

In [5]: x
Out[5]:
['256619580.697790',
 '255721261.271977',
 '336984986.430552',
 '267541556.373619',
 '901048842.193849']

Now they're a list of strings. If we operate on them mathematically, they'll behave like strings, not numbers:

In [6]: x[0] * 2
Out[6]: '256619580.697790256619580.697790'

Controlling how numpy arrays are saved with savetxt

Finally, if you're using numpy.savetxt, and would like to control how the data is output to disk, consider using the fmt parameter instead of manually converting elements of the array to strings.

For example, if we were to do:

np.savetxt('temp.txt', x)

By default, the ascii representation of the array would use scientific notation if it is more compact:

8.702970453168644905e+08
9.991634082796489000e+08
5.032002956810175180e+08
2.382398232565869987e+08
1.868727085152311921e+08

However, we can control that using fmt. Note that it expects the "old-style" % formatting strings:

np.savetxt('temp2.txt', x, fmt='%f')

And we'll get:

870297045.316864
999163408.279649
503200295.681018
238239823.256587
186872708.515231

Answer 2

If you just want to print them without using scientific notation, you can do np.set_printoptions(suppress=True).