Plot sklearn RocCurve-CodePudding

I want to plot my model roc curve but i get error: '<' not supported between instances of 'str' and 'int'

I have numerical and categorical data types.

data_type= dfj.dtypes
data_type
ZAG_LUDNOŚCI                                     int32
GDP_NA_OS                                        int32
DNI POWSZEDNIE                                   int64
PORA_DNIA                                       object
SPECJALNA_BUDOWA_OBIEKTU                         int32
ILOŚĆ_MIESZKAŃCÓW_OBIEKTU                       object
CZY_BUDYNEK_JEST_NORMALNIE_ZAJĘTY                int32
CZY_CZUJNIK_DYMU_JEST_W_OBIEKCIE                object

I deal with them by pipeline:

from sklearn.compose import make_column_selector as selector

numerical_columns_selector = selector(dtype_exclude=object)
categorical_columns_selector = selector(dtype_include=object)

numerical_columns = numerical_columns_selector(data)
categorical_columns = categorical_columns_selector(data)

from sklearn.preprocessing import OneHotEncoder, StandardScaler

categorical_preprocessor = OneHotEncoder(handle_unknown="ignore")
numerical_preprocessor = StandardScaler()

from sklearn.compose import ColumnTransformer

preprocessor = ColumnTransformer([
    ('one-hot-encoder', categorical_preprocessor, categorical_columns),
    ('standard_scaler', numerical_preprocessor, numerical_columns)])

from sklearn.linear_model import LogisticRegression
from sklearn.pipeline import make_pipeline

model = make_pipeline(preprocessor, LogisticRegression(max_iter=1000))

from sklearn import set_config
set_config(display='diagram')
model

Then i learn model and predict:

    from sklearn.model_selection import train_test_split

data_train, data_test, target_train, target_test = train_test_split(
    data, target, random_state=42)
    _ = model.fit(data_train, target_train)
    y_pred=model.predict(data_test)
model.predict(data_test)[:10]
y_pred

at this moment i want to plot a roc curve but i get error message...

from sklearn.metrics import roc_curve
from sklearn.metrics import RocCurveDisplay

y_score = model.decision_function(data_test)

fpr, tpr, _ = roc_curve(data_test, y_score, pos_label=model.classes_[1])
roc_display = RocCurveDisplay(fpr=fpr, tpr=tpr).plot()

The error:

TypeError                                 Traceback (most recent call last)
~\AppData\Local\Temp/ipykernel_20736/4039759825.py in <module>
      4 y_score = model.decision_function(data_test)
      5 
----> 6 fpr, tpr, _ = roc_curve(data_test, y_score, pos_label=model.classes_[1])
      7 roc_display = RocCurveDisplay(fpr=fpr, tpr=tpr).plot()

~\anaconda3\lib\site-packages\sklearn\metrics\_ranking.py in roc_curve(y_true, y_score, pos_label, sample_weight, drop_intermediate)
    977 
    978     """
--> 979     fps, tps, thresholds = _binary_clf_curve(
    980         y_true, y_score, pos_label=pos_label, sample_weight=sample_weight
    981     )

~\anaconda3\lib\site-packages\sklearn\metrics\_ranking.py in _binary_clf_curve(y_true, y_score, pos_label, sample_weight)
    734     """
    735     # Check to make sure y_true is valid
--> 736     y_type = type_of_target(y_true, input_name="y_true")
    737     if not (y_type == "binary" or (y_type == "multiclass" and pos_label is not None)):
    738         raise ValueError("{0} format is not supported".format(y_type))

~\anaconda3\lib\site-packages\sklearn\utils\multiclass.py in type_of_target(y, input_name)
    284         raise ValueError("y cannot be class 'SparseSeries' or 'SparseArray'")
    285 
--> 286     if is_multilabel(y):
    287         return "multilabel-indicator"
    288 

~\anaconda3\lib\site-packages\sklearn\utils\multiclass.py in is_multilabel(y)
    171         )
    172     else:
--> 173         labels = np.unique(y)
    174 
    175         return len(labels) < 3 and (

<__array_function__ internals> in unique(*args, **kwargs)

~\anaconda3\lib\site-packages\numpy\lib\arraysetops.py in unique(ar, return_index, return_inverse, return_counts, axis)
    260     ar = np.asanyarray(ar)
    261     if axis is None:
--> 262         ret = _unique1d(ar, return_index, return_inverse, return_counts)
    263         return _unpack_tuple(ret)
    264 

~\anaconda3\lib\site-packages\numpy\lib\arraysetops.py in _unique1d(ar, return_index, return_inverse, return_counts)
    321         aux = ar[perm]
    322     else:
--> 323         ar.sort()
    324         aux = ar
    325     mask = np.empty(aux.shape, dtype=np.bool_)

TypeError: '<' not supported between instances of 'str' and 'int'

I don't know how to fix it. Thank you for your help! Edit: Target is PRZYPADKOWE_CZY_CELOWE int32 it's look like

0        0
1        0
2        0
3        0
4        1
        ..
12534    0
12535    0
12536    0
12537    0
12538    0
Name: PRZYPADKOWE_CZY_CELOWE, Length: 12539, dtype: int32

CodePudding user response：

The first parameter to roc_curve is supposed to be the true labels, but you are passing the independent variables.