create a numpy array from data on a row

Question

My data file has a sample for each row. Each row is 400 float number. It's a 20x20 image on a single line. I have to write a numpy array with dimensions (number of row, 20, 20, 1). Last dimension is the value (the float number in the file).

I tried something like:

X1=[]
for x in range (1,nrow+1):
    for a in range (1,21):
        for b in range (1,21):
           index = a*b-1
           X1.append((x,a,b,X[x,index]))
X = np.array(X1)

but I know this is wrong.

EDIT:

Maybe this is the solution:

X1=[]
for x in range (1,Nsamples+1):
    for a in range (1,21):
        for b in range (1,21):
           index = a*b-1
           X1.append((X[x,index]))
           #X1.append((X[x,index], x))
X = np.array(X1)
X = X.reshape(Nsamples,20,20,1)

If you know this is wrong? Why do you think that? Please provide any exception you encountered while running the code? Also what is nrow — Devesh Kumar Singh
– Devesh Kumar Singh, Commented May 6, 2019 at 11:22
I know it's wrong because with one row the shape of the X array is (400, 4) and i would like to get (1,20,20,1). nrow is the number of row in the file. Each row is 400 float number. — a.p.
– a.p., Commented May 6, 2019 at 11:26
No. Sorry but I am a beginner. nrow is not fixed. Presently it's 1 just for testing, Each row has 400 float number that should go in a 20x20 matrix. — a.p.
– a.p., Commented May 6, 2019 at 11:30
If the dimensions have to include the "number of row", that means the dimension is variable, since there are 20 rows, each with a different number. So I'm unclear what your actual intention is. Perhaps it should be "numbers of rows"? — 9769953
– 9769953, Commented May 6, 2019 at 11:32

nassim · Accepted Answer · 2019-05-06 11:41:18Z

1

try using reshape

X = X.reshape(X.shape[0],20,20,1)

it will give you a numpy array of reshaped images with same number of lines

answered May 6, 2019 at 11:41

nassim

1,5811 gold badge16 silver badges28 bronze badges

Sign up to request clarification or add additional context in comments.

9 Comments

a.p. Over a year ago

ValueError: cannot reshape array of size 1600 into shape (400,20,20,1).

a.p. Over a year ago

X.shape[0] is 400

nassim Over a year ago

what is X.shape ?

a.p. Over a year ago

X.shape is (400, 4) so I can't reshape to (1,20,20,1)

nassim Over a year ago

your 400 floats are in a column not in a row

|

a.p. · Accepted Answer · 2019-05-06 12:19:17Z

0

This seems to be the correct answer, thanks to people who helped.

X1=[]
for x in range (1,Nsamples+1):
    for a in range (1,21):
        for b in range (1,21):
           index = a*b-1
           X1.append((X[x,index]))
X = np.array(X1)
X = X.reshape(Nsamples,20,20,1)

answered May 6, 2019 at 12:19

a.p.

14 bronze badges

Collectives™ on Stack Overflow

create a numpy array from data on a row

2 Answers 2

9 Comments

Comments

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

9 Comments

Comments

Related