AWS SageMaker MasterClass

Explore how machine learning enables computers to learn from data and predict outcomes without programming, using supervised, unsupervised, semi supervised, and reinforcement learning, with techniques like regression, classification, and segmentation.

Explore the complete machine learning lifecycle from data gathering and preparation to model training, evaluation, and deployment, including data exploration, preprocessing, and selecting appropriate techniques.

Explore a practical regression workflow on an insurance dataset, encoding categorical features, and comparing linear regression, decision tree, and random forest to predict charges.

Compare logistic regression, decision tree, and random forest classifiers on a diabetes dataset, performing data import, train-test split, model training, and accuracy evaluation.

Discover how a machine learning pipeline automates data ingestion, cleaning, feature engineering, modeling, and deployment through modular independent components for scalable machine learning tasks.

Explore hyperparameter tuning in machine learning with grid search CV, distinguishing parameters from hyperparameters and using a param grid and cross-validation to find best values in a practical svc example.

Explore essential machine learning evaluation metrics for classification and regression, including confusion matrix, accuracy, precision, recall, and mean absolute error and mean squared error.

Explore cloud computing fundamentals, on-demand data storage and services, and the three service models: SaaS, PaaS, and IaaS, plus providers and data centers.

Discover how AWS, Amazon Web Services, provides 200+ fully featured cloud services with regional availability zones, scalable on-demand computing and storage, and a secure, cost-effective infrastructure.

Explore AWS services such as S3 object storage, Amazon Relational Database Service, IAM roles, EC2 compute, and SageMaker for cloud machine learning.

Explore AWS SageMaker, a fully managed machine learning workflow platform, covering data labeling, preprocessing, model training, evaluation, hyperparameter tuning, and deployment in production.

Set up an AWS SageMaker notebook instance, upload the diabetes dataset, train a random forest classifier in a Jupyter notebook, achieve about 78% accuracy, and terminate resources to avoid costs.

Learn to choose the right supervised learning algorithms in AWS SageMaker, including binary and multiclass classification and regression, and explore built-in models like linear learner and extreme boost algorithms.

Demonstrates a practical breast cancer prediction with AWS SageMaker linear learner: load data to S3, train a model in a notebook, deploy a hosted endpoint, and evaluate accuracy.

Explore no-code machine learning with AWS SageMaker Canvas, automatically cleaning data, pre-processing, selecting models, and generating single or bulk predictions from uploaded datasets.

Explore the AWS SageMaker marketplace to access pre-trained model packages, try product demos, and evaluate ready-made object detection models like yolov3 for various tasks.