Transform list of strings (indexes) to Numpy Array

Question

I have a list where every value is presented only one, and I have another list with the desired order of a tokenized numpy array.

For instance:

sorted_values = ['a', 'b', 'c', 'd', 'e', 'f', 'g']
a = ['a', 'c', 'g']
b = ['e']

I want to convert a and b (in a efficient) to a numpy array like this:

at = [1,0,1,0,0,0,1]
bt = [0,0,0,0,1,0,0]

Is there any efficient way to do this?

Divakar · Accepted Answer · 2017-05-02 19:38:54Z

3

Using the fact that the first array is already sorted, we could employ np.searchsorted for efficiency -

at = np.zeros(len(sorted_values), dtype=int)
bt = at.copy()
at[np.searchsorted(sorted_values, a)] = 1
bt[np.searchsorted(sorted_values, b)] = 1

answered May 2, 2017 at 19:38

Divakar

222k19 gold badges273 silver badges374 bronze badges

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

Comments

akuiper · Accepted Answer · 2017-05-02 19:37:57Z

2

You can use np.in1d:

np.in1d(sorted_values, a).astype(int)
#array([1, 0, 1, 0, 0, 0, 1])

np.in1d(sorted_values, b).astype(int)
#array([0, 0, 0, 0, 1, 0, 0])

answered May 2, 2017 at 19:37

akuiper

216k33 gold badges362 silver badges379 bronze badges