Using string as parameter name

Question

I have the following functions:

def train(data,**kwargs):
    train_model(data,**kwargs)

def test_train_parameters(data1,data2,parameter_to_test, parameter_value):
    train(data1,**kwargs)
    train(data2,**kwargs)

Train function has some optional parameters, such as gamma, lambda, num_rounds and so on. Train with data1 would be called without any modified parameter, but train with data2 would be called with a parameter modified and it's value. For example, let's say I want a gamma = 5, I would code:

test_train_parameters(parameter_to_test = 'gamma', parameter_value = 5)

And the function would be called as follow:

train(data1)
train(data2, gamma = 5)

Thus, this can not be done just calling the parameter as test_train_parameters(gamma = 5)because it would interfere with the first train.

I have been searching on google but I have been unable to find something that fits my case (I have found eval, getattr, passing list... but those are for other things). How could I make it?

What if you pass two dictionaries (one specifically for each train) - therefore parameters could be passed explicitly. — Jeremy H
– Jeremy H, Commented Aug 8, 2019 at 16:09
You can just do train(data2,**kwargs, **{parameter_to_test: parameter_value}). Or if parameter_to_test may appear in kwargs, do train(data1,**kwargs); kwargs[parameter_to_test] = parameter_value; train(data2,**kwargs). — javidcf
– javidcf, Commented Aug 8, 2019 at 16:10

Alexandre Senges · Accepted Answer · 2019-08-08 16:16:37Z

4

In python, kwargs are just dictionaries, so you can create a dictionary in your function with your arguments:

def test_train_parameters(data1,data2,parameter_to_test, parameter_value):
    train(data1)
    kwargs = {parameter_to_test: parameter_value}
    train(data2,**kwargs)

answered Aug 8, 2019 at 16:16

Alexandre Senges

1,6091 gold badge15 silver badges23 bronze badges

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

1 Comment

Javier Lopez Tomas Over a year ago

What if I already have other parameters passing as kwargs? Making kwargs[parameter_to_test] = parameter_value would be enough to add this one to the others?

zslim · Accepted Answer · 2019-08-08 16:11:03Z

1

An if with every possible parameter? Like

def test_train_parameters(data1,data2,parameter_to_test, parameter_value):
    train(data1)
    if parameter_to_test == 'alpha':
        train(data2, alpha=parameter_value)
    elif parameter_to_test == 'beta':
        train(data2, beta=parameter_value)
    elif parameter_to_test == 'gamma':
        train(data2, gamma=parameter_value)

I know it's not pleasing to the eye but it's a simple way to do it.

answered Aug 8, 2019 at 16:11

zslim

4911 gold badge8 silver badges15 bronze badges

1 Comment

Javier Lopez Tomas Over a year ago

Well, there are like 50 parameters that can be modified, so it will be very tedious to make such a solution

Collectives™ on Stack Overflow

Using string as parameter name

2 Answers 2

1 Comment

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related