Shortest Syntax To Use numpy 1d-array As sklearn X

Question

I often have two numpy 1d arrays, x and y, and would like to perform some quick sklearn fitting + prediction using them.

 import numpy as np
 from sklearn import linear_model

 # This is an example for the 1d aspect - it's obtained from something else.
 x = np.array([1, 3, 2, ...]) 
 y = np.array([12, 32, 4, ...])

Now I'd like to do something like

 linear_model.LinearRegression().fit(x, y)...

The problem is that it expects an X which is a 2d column array. For this reason, I usually feed it

 x.reshape((len(x), 1))

which I find cumbersome and hard to read.

Is there some shorter way to transform a 1d array to a 2d column array (or, alternatively, get sklearn to accept 1d arrays)?

Answer 1

You can slice your array, creating a newaxis:

x[:, None]

This:

>>> x = np.arange(5)
>>> x[:, None]
array([[0],
       [1],
       [2],
       [3],
       [4]])

Is equivalent to:

>>> x.reshape(len(x), 1)
array([[0],
       [1],
       [2],
       [3],
       [4]])

If you find it more readable, you can use a transposed matrix:

np.matrix(x).T

If you want an array:

np.matrix(x).T.A