I have a dictionary
mydict = {'jon': 12, 'alex': 17, 'jane': 13}
and I want to create a np.array which contains the values 12, 17, 13, but sorted by another array
sortby = np.array(['jon', 'jane', 'alex'])
which should yield the output as
sorted_array = np.array([12, 13, 17])
Any approaches that are more efficient than looping through the sortby array like below?
sorted_array = []
for vals in sortby:
sorted_array.append(mydict[vals])
return np.array(sorted_array)
Use list comprehension,
In [100]: np.array([mydict[i] for i in sortby])
Out[100]: array([12, 13, 17])
Edit:
Execution timings, To make clear for mohammad and Moses Discussions
In [119]: def test_loop():
sorted_array = []
for vals in sortby:
sorted_array.append(mydict[vals])
return np.array(sorted_array)
.....:
In [120]: def test_list_compres():
return np.array([mydict[i] for i in sortby])
.....:
In [121]: %timeit test_list_compres
10000000 loops, best of 3: 20 ns per loop
In [122]: %timeit test_loop
10000000 loops, best of 3: 21.3 ns per loop
In [123]: %timeit test_list_compres
10000000 loops, best of 3: 20.1 ns per loop
In [124]: %timeit test_loop
10000000 loops, best of 3: 21.9 ns per loop
It's a marginal difference but it will make a significant change with huge entries.
