Numpy array operation based on user input

Question

I have a 2D numpy array. Lets say:

abc = np.arange(40).reshape(10,4)

Now I am accepting a user input string variable 'mm'. This variable determines what I want to do with the array. It has the following 4 possible values:

'min' : Returns a column with all the minimum values

'max' : Returns a column with all the maximum values

'mean': Returns a column with all the mean values

'time' : Returns a column with the value at a particular column. It also needs an additional input from the user; variable 'idx' with the column number.

I have included the first three possibilities by defining a dictionary variable that chooses the relevant function.

dict = {'mean':np.mean, 'max':np.max, 'min':np.min}
fn = dict[mm]

Therefore, my output variable will be:

op = fn(abc,axis=1)

But I am stuck when I try to include the 4th possibility of slicing a particular column based on user input. Is it possible to define any numpy function within the dictionary 'dict' that can include this operation as well?

I would prefer not to use an if-else condition, as it makes the code too long. The actual variables are much larger and more complicated to handle.

The expected outcome when the user variable 'mm' has the different values:

op_min = np.min(abc,axis=1)
op_max = np.max(abc,axis=1)
op_mean = np.mean(abc,axis=1)
op_time = abc[:,2]  # assuming the user input for idx is 2

Yes, that looks like a function that does the task which I wanted. However, it has an additional argument compared to np.max, np.min etc. But I think it can still be overridden in the same manner that I did for the answer posted by me. Thanks a lot anyway! — joseph praful
– joseph praful, Commented Mar 26, 2019 at 15:48

Ehsan · Accepted Answer · 2019-03-26 13:36:16Z

2

If you can transpose the matrix , this function can select a row:

abc = np.arange(40).reshape(10,4)
row_num = 2
f = operator.itemgetter(row_num)
f(abc)

Also you can define one line function by lambda for selecting a column:

f = lambda x: x[:,column_num]
f(abc)

answered Mar 26, 2019 at 13:36

Ehsan

732 silver badges6 bronze badges

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

1 Comment

joseph praful Over a year ago

Yes I can transpose the matrix. And this seems like exactly the kind of function I was looking for. But I still have one problem, how to go around the axis parameter in np.mean, np.max etc. Coz itemgetter does not accept any axis keyword argument, but the other functions require this keyword.

joseph praful · Accepted Answer · 2019-03-26 15:42:52Z

0

Okay, I found a solution. I transposed the matrix as suggested by user3371603 and used the function operator.getitem. Then I used an index variable (idx) with a default value of 0. So,

import operator
dict = {'mean':np.mean, 'max':np.max, 'min':np.min, 'time':operator.getitem}
fn = dict[mm]

idx = 0
if mm == 'time': idx=2     # or user input

cba = np.swapaxes(abc,0,1)
op = fn(cba, idx)

Special thanks to user3371603!

answered Mar 26, 2019 at 15:42

joseph praful

1734 silver badges16 bronze badges

Collectives™ on Stack Overflow

Numpy array operation based on user input

2 Answers 2

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related