Noções Básicas de MLOps

uv venv && source .venv/bin/activate
uv pip install "mlflow>=2.20" "scikit-learn>=1.5" "fastapi>=0.115" "joblib"

import mlflow
import mlflow.sklearn
from sklearn.ensemble import RandomForestClassifier
 
with mlflow.start_run():
    clf = RandomForestClassifier(n_estimators=100, random_state=42)
    clf.fit(X_train, y_train)
    mlflow.log_param("n_estimators", 100)
    mlflow.log_metric("accuracy", clf.score(X_test, y_test))
    mlflow.sklearn.log_model(clf, "model")

import joblib
joblib.dump(pipeline, "models/churn_v1.joblib")

pipeline = joblib.load("models/churn_v1.joblib")
prediction = pipeline.predict(new_data)

import hashlib
data_hash = hashlib.sha256(open("train.csv", "rb").read()).hexdigest()[:12]

# train.py - execute com: python train.py --data train.csv
def main(data_path: str) -> None:
    ...
if __name__ == "__main__":
    main()

from fastapi import FastAPI
app = FastAPI()
@app.post("/predict")
def predict(features: dict):
    return {"prediction": int(model.predict([list(features.values())])[0])}

def test_predict():
    assert model.predict(sample_input).shape == (1,)

dvc add data/train.csv
git add data/train.csv.dvc

client = mlflow.tracking.MlflowClient()
client.transition_model_version_stage("churn", version=3, stage="Staging")

import logging
logging.info(f"prediction={pred} features_hash={hash}")