End-to-End-Cancer-Classification-using-MLflow-DVC/example.py

import os
import warnings
import sys

import pandas as pd
import numpy as np
from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score
from sklearn.model_selection import train_test_split
from sklearn.linear_model import ElasticNet
from urllib.parse import urlparse
import mlflow
from mlflow.models.signature import infer_signature
import mlflow.sklearn

import logging 

logging.basicConfig(level = logging.WARN)
logger = logging.getLogger(__name__)


def eval_metrics(actual, pred):
    rmse = np.sqrt(mean_squared_error(actual, pred))
    mae = mean_absolute_error(actual, pred)
    r2 = r2_score(actual, pred)
    return rmse, mae, r2


if __name__=="__main__":
    warnings.filterwarnings("ignore")
    np.random.seed(40)


    # Read teh wine-quality csv file from thr url
    csv_url = ""
    try:
        data =  pd.read_csv(csv_url, sep=";")
    except Exception as e:
        logger.exception("Unable to download csv")

    # Split the data into trainin g and test test sets
    train, test = train_test_split(data)


    # The predicted column is quality which is a scaler from [3,9]
    train_x = train.drop(["quality"], axis = 1)
    test_x = test.drop(["quality"], axis = 1)
    train_y = train["quality"]
    test_y = test["quality"]


    alpha = float(sys.argv[1]) if len(sys.argv)> 1 else 0.5
    l1_ratio = float(sys.argv[2]) if len(sys.argv)> 2 else 0.5

    with mlflow.start_run():
        lr = ElasticNet(alpha=alpha, l1_ratio=l1_ratio,random_state=42)
        lr.fit(train_x, train_y)

        predicted_qualities = lr.predict(test_x)

        (rmse, mae, r2) = eval_metrics(test_y, predicted_qualities)

        print("Elasticnet model (alpha = {:f}, l1_ratio = {:f}):".format(alpha, l1_ratio))
        print(" RMSE: %s" % rmse)
        print(" MAE: %s" % mae)
        print(" R2: %s" % r2)

        mlflow.log_param("alpha", alpha)
        mlflow.log_param("l1_ratio", l1_ratio)
        mlflow.log_metric("rmse", rmse)
        mlflow.log_metric("mae", mae)
        mlflow.log_metric("r2", r2)

        predictions = lr.predict(train_x)
        signature = infer_signature(train_x, predictions)

        tracking_url_type_store = urlparse(mlflow.get_tracking_uri()).scheme

        # Model registry does not work with file store
        if tracking_url_type_store != "file":
            mlflow.sklearn.log_model(
                lr, "model", registered_model_name="ElasticnetWineModel", signature=signature
            )

        else:
            mlflow.sklearn.log_model(lr, "model", signature=signature)