[Python] ValueError when fitting scikit-learn pipeline with OneHotEncoder and...

I’m trying to build a simple ML pipeline in Python using scikit-learn.
My dataset has both categorical and numerical features. I want to apply OneHotEncoder to the categorical columns and then fit a LogisticRegression model.

Here’s my code:

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import OneHotEncoder
from sklearn.compose import ColumnTransformer
from sklearn.pipeline import Pipeline
from sklearn.linear_model import LogisticRegression

data = pd.DataFrame({
"age": [25, 32, 40, 29],
"city": ["Delhi", "Mumbai", "Delhi", "Bangalore"],
"purchased": [0, 1, 0, 1]
})

X = data[["age", "city"]]
y = data["purchased"]

categorical_features = ["city"]
numeric_features = ["age"]

preprocessor = ColumnTransformer(
transformers=[
("cat", OneHotEncoder(), categorical_features),
("num", "passthrough", numeric_features)
]
)

pipeline = Pipeline(steps=[
("preprocessor", preprocessor),
("model", LogisticRegression())
])

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25)

pipeline.fit(X_train, y_train) # <-- Error here


When I run this, I get:

ValueError: could not convert string to float: 'Delhi'


But I thought OneHotEncoder should handle categorical strings.


Checked scikit-learn version (1.3.0).

Tried using handle_unknown='ignore' in OneHotEncoder.

Verified that city column is object dtype.

Why am I still getting this error, and how can I correctly fit this pipeline with categorical and numeric features?

Continue reading...