将 2D 数组更改为 3D 数组

Question

I have an array that has 8450 rows and 16 columns. I want to feed these data points into an RNN with each 50 points being an entry. So 0-49 is z=0, and 1-50 is z=1 and so forth. The columns need to remain unchanged so that I can still have the same data in each z axis entry. So basically I am taking every chunk of 50 points and moving it into a third axis. Is there simple way to do this python? I tried the reshape but I may not have been doing it correctly. Currently the data is in a pandas dataframe.

points = 50
for i in range(len(data_prepped_dataframe)-points):
     x_data = data_prepped_dataframe.iloc[i:i+points,:]

So far I have this but all this does is give me the last 50 points in the data set. I tried adding indexes to the x_data term but that threw an error

I tried

x_data[:,:,i] = data_prepped_dataframe.iloc[i:i+points,:]

but the error said x_data wasn't defined.

Answer 1

If you change the dataframe to an array using np.array() and then add .copy() at the end this will output the 3D array. Specifically in this case a (8400 x 50 x 16) array