Operation vectorization in numpy

Question

I am given an numpy array of dates(date is a number from 0 to 371).

dates = [234, 12, 343, ...]

and and empty array date_counter (of size 372).

date_counter = [0, 0, ..., 0]

For every date from dates I would like to increment date_counter . How can I make this operation vectorized?

Answer 1

Count the number of occurrences for value and then update the date counter.

For instance, to get the increment:

unique, counts = np.unique(dates, return_counts=True)

And then:

date_counter[unique] += counts

Answer 2

I think np.histogram is your friend here. It does exactly what you're asking: counts items and puts them into "buckets". A smaller (human readable) example:

In [9]: vals = [9, 0, 1, 2, 2, 9, 3, 7, 8, 9, 9]

In [10]: date_counter, _ = np.histogram(vals, bins=np.arange(11)) # 11 -> 373 for your case

In [11]: date_counter
Out[11]: array([1, 1, 2, 1, 0, 0, 0, 1, 1, 4], dtype=int64)

Each index corresponds to the number of items with value equal to that of said index (ie the last number in date_counter is 4 , located at index 9 , meaning there are four 9 s) as desired.

HTH.