将字典值从1D数组变形为2D数组

Question

I want to calculate cosine similarity for words in two dictionaries. The words are keys, arrays are values. In order to do this, I need to convert them to 2D array first, I did some research and found x = x.reshape(1,-1) I successfully converted it with a single value in dictionary, however, I don't know how to convert the whole dictionary's values by using for loops.

The data

D
{'A': array([ 4.80625004e-01, -1.40245005e-01, -9.99999046e-03]),
'B': array([-0.46553   , -0.1519755 ,  0.41836]),
'C': array([0.0090175 ,  0.05817001, -0.09712502])}

D2 (same format as D)
{'D': ([8.11059952e-01,  6.84859991e-01,  1.01619996e-01]),
'E':([-0.82868   ,  0.49513   ,  0.67581]),
'F':([-0.17093   ,  0.88746   ,  0.0931135])}

I tried

for i in D2.items():
    D2[i] = D2[i].reshape(1, -1) #Error on this line

but received error: TypeError: unhashable type: 'numpy.ndarray'

Some advice please? thank you in advanced!

Answer 1

Dict.items() returns tuples of the key, value pairs. Therefore in your case i is a tuple of the key and the value. Try:

for i, value in D2.items():
    D2[i] = value.reshape(1, -1)

Answer 2

try this

from numpy import array

d = dict({'A': array([ 4.80625004e-01, -1.40245005e-01, -9.99999046e-03]),
'B': array([-0.46553   , -0.1519755 ,  0.41836]),
'C': array([0.0090175 ,  0.05817001, -0.09712502])})

reshaped_arrays = dict()
    for x, j in d.items():
        dd[x]=j.reshape(1,-1)