pip install scikit-learn skl2onnx onnxruntime

from sklearn.ensemble import RandomForestClassifier
from sklearn.datasets import load_iris
from skl2onnx import convert_sklearn
from skl2onnx.common.data_types import FloatTensorType
import numpy as np

# Train a model
data = load_iris()
X, y = data.data, data.target
model = RandomForestClassifier(n_estimators=10, random_state=42)
model.fit(X, y)

# Define input type (shape: [batch_size, n_features])
initial_type = [('float_input', FloatTensorType([None, 4]))]

# Convert to ONNX
onnx_model = convert_sklearn(
    model,
    initial_types=initial_type,
    target_opset=14
)

# Save the model
with open("rf_classifier.onnx", "wb") as f:
    f.write(onnx_model.SerializeToString())

from sklearn.linear_model import LinearRegression
from skl2onnx import convert_sklearn
from skl2onnx.common.data_types import FloatTensorType
import numpy as np

# Create and train model
X = np.random.randn(100, 5).astype(np.float32)
y = X.sum(axis=1) + np.random.randn(100) * 0.1

model = LinearRegression()
model.fit(X, y)

# Convert to ONNX
initial_type = [('float_input', FloatTensorType([None, 5]))]
onnx_model = convert_sklearn(
    model,
    initial_types=initial_type,
    target_opset=14
)

# Save
with open("linear_regression.onnx", "wb") as f:
    f.write(onnx_model.SerializeToString())

from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.datasets import make_classification
from skl2onnx import convert_sklearn
from skl2onnx.common.data_types import FloatTensorType

# Create pipeline
X, y = make_classification(n_samples=1000, n_features=20, random_state=42)

pipeline = Pipeline([
    ('scaler', StandardScaler()),
    ('classifier', GradientBoostingClassifier(n_estimators=50, random_state=42))
])

pipeline.fit(X, y)

# Convert entire pipeline
initial_type = [('float_input', FloatTensorType([None, 20]))]
onnx_model = convert_sklearn(
    pipeline,
    initial_types=initial_type,
    target_opset=14
)

with open("pipeline.onnx", "wb") as f:
    f.write(onnx_model.SerializeToString())

from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.ensemble import RandomForestClassifier
from skl2onnx import convert_sklearn
from skl2onnx.common.data_types import FloatTensorType, Int64TensorType
import numpy as np

# Prepare data
X_numeric = np.random.randn(100, 3).astype(np.float32)
X_categorical = np.random.randint(0, 5, size=(100, 2))
X = np.hstack([X_numeric, X_categorical])
y = np.random.randint(0, 2, size=100)

# Create pipeline with mixed types
preprocessor = ColumnTransformer(
    transformers=[
        ('num', StandardScaler(), [0, 1, 2]),
        ('cat', OneHotEncoder(), [3, 4])
    ])

model = Pipeline([
    ('preprocessor', preprocessor),
    ('classifier', RandomForestClassifier(n_estimators=10))
])

model.fit(X, y)

# Define mixed input types
initial_type = [
    ('numeric_input', FloatTensorType([None, 3])),
    ('categorical_input', Int64TensorType([None, 2]))
]

onnx_model = convert_sklearn(
    model,
    initial_types=initial_type,
    target_opset=14
)

from skl2onnx import convert_sklearn
from skl2onnx.common.data_types import FloatTensorType
from sklearn.ensemble import RandomForestClassifier

# Train model
model = RandomForestClassifier(n_estimators=100)
model.fit(X_train, y_train)

# Convert with options
initial_type = [('float_input', FloatTensorType([None, n_features]))]
options = {
    'zipmap': False,  # Disable ZipMap for classification
    'nocl': False,    # Keep class labels
    'raw_scores': True  # Output raw scores instead of probabilities
}

onnx_model = convert_sklearn(
    model,
    initial_types=initial_type,
    target_opset=14,
    options=options
)

import onnxruntime as ort
import numpy as np

# Load ONNX model
session = ort.InferenceSession("rf_classifier.onnx")

# Prepare input
test_input = np.array([[5.1, 3.5, 1.4, 0.2]], dtype=np.float32)

# Get input/output names
input_name = session.get_inputs()[0].name
output_names = [output.name for output in session.get_outputs()]

# Run inference
results = session.run(output_names, {input_name: test_input})

# Parse results
label = results[0]  # Predicted class label
probabilities = results[1]  # Class probabilities

print(f"Predicted class: {label[0]}")
print(f"Probabilities: {probabilities}")

import numpy as np
from sklearn.ensemble import RandomForestClassifier
import onnxruntime as ort

# Original sklearn prediction
X_test = np.random.randn(10, 4).astype(np.float32)
sklearn_pred = model.predict(X_test)
sklearn_proba = model.predict_proba(X_test)

# ONNX prediction
session = ort.InferenceSession("rf_classifier.onnx")
input_name = session.get_inputs()[0].name
onnx_results = session.run(None, {input_name: X_test})

# Compare
labels_match = np.array_equal(sklearn_pred, onnx_results[0])
proba_close = np.allclose(sklearn_proba, onnx_results[1], rtol=1e-5)

if labels_match and proba_close:
    print("✓ Validation successful")
else:
    print("✗ Validation failed")
    print(f"Labels match: {labels_match}")
    print(f"Probabilities close: {proba_close}")

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LogisticRegression
from sklearn.pipeline import Pipeline
from skl2onnx import convert_sklearn
from skl2onnx.common.data_types import StringTensorType

# Sample text data
texts = [
    "This is a positive example",
    "This is a negative example",
    "Another positive text",
    "Another negative text"
]
labels = [1, 0, 1, 0]

# Create and train pipeline
pipeline = Pipeline([
    ('tfidf', TfidfVectorizer(max_features=100)),
    ('classifier', LogisticRegression())
])

pipeline.fit(texts, labels)

# Convert to ONNX
initial_type = [('input', StringTensorType([None, 1]))]
onnx_model = convert_sklearn(
    pipeline,
    initial_types=initial_type,
    target_opset=14
)

with open("text_classifier.onnx", "wb") as f:
    f.write(onnx_model.SerializeToString())

from sklearn.impute import SimpleImputer
from sklearn.ensemble import RandomForestRegressor
from sklearn.pipeline import Pipeline
from skl2onnx import convert_sklearn
from skl2onnx.common.data_types import FloatTensorType

# Pipeline with imputation
pipeline = Pipeline([
    ('imputer', SimpleImputer(strategy='mean')),
    ('regressor', RandomForestRegressor(n_estimators=10))
])

pipeline.fit(X_train, y_train)

# Convert
initial_type = [('float_input', FloatTensorType([None, n_features]))]
onnx_model = convert_sklearn(
    pipeline,
    initial_types=initial_type,
    target_opset=14
)

pip install --upgrade skl2onnx

import time
import numpy as np

# Benchmark sklearn
X_test = np.random.randn(1000, 4).astype(np.float32)

start = time.time()
for _ in range(100):
    _ = model.predict(X_test)
sklearn_time = time.time() - start

# Benchmark ONNX Runtime
session = ort.InferenceSession("rf_classifier.onnx")
input_name = session.get_inputs()[0].name

start = time.time()
for _ in range(100):
    _ = session.run(None, {input_name: X_test})
onnx_time = time.time() - start

print(f"scikit-learn: {sklearn_time:.3f}s")
print(f"ONNX Runtime: {onnx_time:.3f}s")
print(f"Speedup: {sklearn_time/onnx_time:.2f}x")