CODEWITHSUNDEEP
pythonnumpy
Elkan
Elkan

Reputation: 616

How can I identify numpy's integer type?

What is the simplest way to identify whether a data point in numpy's array is an integer? Currently I use numpy.dtype(x[i]).type to return the type of the element i of array x and then 

`if numpy.dtype(x[i]).type is numpy.int*`

to achieve this, where * can be 8, 32 or 64. But it may also return uint, thus this if way can return False. I wonder whether there exists a simple way to identify whether it is an integer, regardless of the exact int type is. And how about float?

Upvotes: 1

Views: 905

Answers (3)

ewcz
ewcz

Reputation: 13087

this might help (this effectively tests for a signed integer, in a similar fashion 'u' would be unsigned integer, etc.):

x[i].dtype.kind == 'i'

Upvotes: 3

Eric
Eric

Reputation: 97571

Use either:

 issubclass(x[i].dtype.type, np.integer)

or

 np.issubdtype(x[i].dtype, np.integer)

Your code of np.dtype(x) is unlikely to do what you want - that doesn't get the dtype of x, but tries to interpret x as a description of a new dtype. If you have a possibly non-numpy object you want a dtype of, you can use np.asanyarray(x).dtype

Upvotes: 1

sai
sai

Reputation: 462

You can use:

issubdtype(var, type)

Usage:

numpy.issubdtype(var_to_check, np.integer)

More information here How to determine if a number is any type of int (core or numpy, signed or not)?

Upvotes: 4

Related Questions