Selectbox `format_func` with classes

Using Streamlit
1

I have the following code where I would like the user to choose 3 machine learning models. I want the UI to look nice, so instead of having to select the class name, I want to format all the options like so:

records = [
        {'model':LogisticRegression(),'name':'Logistic Regression'},
        {'model':DecisionTreeClassifier(),'name':'Decision Tree Classifier'}
]

with st.sidebar:
    with st.form('Form1'):
        selection1 = st.selectbox(
            'Model 1', options=records,
            format_func=lambda record: f'{record["name"]}'
        )
        model1 = selection1.get('model')

        selection2 = st.selectbox(
            'Model 2', options=records,
            format_func=lambda record: f'{record["name"]}'
        )
        model2 = selection2.get('model')

        selection3 = st.selectbox(
            'Model 3', options=records,
            format_func=lambda record: f'{record["name"]}'
        )
        model3 = selection3.get('model')

        submitted = st.form_submit_button('Run')

if not submitted:
    st.write('foo')

else:
    st.write(
        model1,model2,model3
    )

However, I get an ValueError when trying this

ValueError: {'model': LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,
                   intercept_scaling=1, l1_ratio=None, max_iter=100,
                   multi_class='warn', n_jobs=None, penalty='l2',
                   random_state=None, solver='warn', tol=0.0001, verbose=0,
                   warm_start=False), 'name': 'Logistic Regression'} is not in iterable

I’ve done this before except I didn’t uses classes. For example, If I change everything to a string, like so

records = [
        {'model':'LogisticRegression()','name':'Logistic Regression'},
        {'model':'DecisionTreeClassifier()','name':'Decision Tree Classifier'}
]

The code runs, but I obviously can’t use those models since they are just strings. So my question is, how can I format classes within a selectbox? Again, I don’t want

LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,
                   intercept_scaling=1, l1_ratio=None, max_iter=100,
                   multi_class='warn', n_jobs=None, penalty='l2',
                   random_state=None, solver='warn', tol=0.0001, verbose=0,
                   warm_start=False)

to be an option in the selectbox, but rather just

Logistic Regression

and then have the selection assigned to the class.

2

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.