numpy arrays 的高效 for 循环

Question

Data_File I have a data frame of size [8192x29], i want to calculate the fft of each column but before that i need to reshape every column (8192 x 1) to (64 x 128). So for one column it goes like this

data_1 = np.array(data_all.iloc[:,16])
data_1 = np.reshape(data_1,(64,128))
data_standarized = preprocessing.scale(data_1)
fft = fftpack.fft(data_standarized)

Now i have to apply the same for all the data and hence "for loop" for all the columns and calculates the fft separatel. I can't seem to think an efficient way to do that. my for loop is as follows

data_k = np.zeros((64,3712))
for i in range(0,data_size[1]):
        data_k[0:63, i:i+127]= data_all[:,i].reshape(64,128)

Then i need another for loop to get the 64 x 128 matrix one by one and apply fft. In MATLAB if we have the structures we can do it with one for loop and every cell contains the output. Can we do something like this in Python?

Answer 1

fftpack.fft will be applied to the last axis by default but you can specify any other axis as well:

data_new = data.reshape(64, 128, 29)
data_fft = fftpack.fft(data_new, axis=1)

Then you can reshape again if you like: data_fft.reshape(64, -1) .