{"id":5647111,"date":"2023-01-12T14:27:18","date_gmt":"2023-01-12T19:27:18","guid":{"rendered":"https:\/\/lightning.ai\/pages\/?p=5647111"},"modified":"2023-03-07T17:09:34","modified_gmt":"2023-03-07T22:09:34","slug":"train-scikit-learn-models","status":"publish","type":"post","link":"https:\/\/lightning.ai\/pages\/community\/tutorial\/train-scikit-learn-models\/","title":{"rendered":"Train Scikit-learn Models on the Cloud"},"content":{"rendered":"

Key takeaways<\/h3> Train Scikit-learn models on the cloud. <\/div>\n

The scale of machine learning datasets can often become large enough that migrating to a cloud provider for training, storing checkpoints, and logs becomes necessary. Making this migration, however, brings with it a host of attendant cloud infrastructure overhead that can be new to (and difficult for!) machine learning engineering teams.<\/p>\n

In this blog post, we’ll show you how to train a Scikit-learn model with the Lightning framework. Then, we’ll explore how you can migrate your existing training script to train with Lightning.<\/p>\n

Building a Scikit-learn training component<\/h2>\n

We’ll use the Iris flower dataset<\/span><\/a> composed of three kinds of irises with different sepal and petal lengths. The chart below shows the distribution of flowers by their sepal width and height. The three color dots represent the category of Iris plants.<\/p>\n

\"\"

Distribution of Iris flower dataset. Source: Scikit-Learn<\/p><\/div>\n

Scikit-learn already offers a convenient prebuilt function to load the iris dataset. For this tutorial, we’ll load the data using Scikit-learn and train a decision tree classifier.<\/p>\n

Below is a simple code example for loading and using the dataset to train a decision tree classification model without Lightning. While this is the method in which Scikit-learn is typically used, combining it with cloud services for large datasets or deployment can be difficult.<\/p>\n

\n\nfrom sklearn.datasets import load_iris
\nfrom sklearn import tree
\nfrom sklearn.model_selection import train_test_split\n\n# Load the iris dataset
\niris = load_iris()
\nX, y = iris.data, iris.target\n\n# Split the dataset into training and test set
\nX_train, X_test, y_train, y_test = train_test_split(X, y)\n\n# initialize a Decision tree model
\nclf = tree.DecisionTreeClassifier()\n\n# train the model
\nclf = clf.fit(X_train, y_train)\n\n# check accuracy
\nprint(f\"train accuracy: {clf.score(X_train, y_train)}\")
\nprint(f\"test accuracy: {clf.score(X_test, y_test)}\")\n\n<\/code>