Build a Titanic survival model with visual tools

This 2019 session used the Titanic dataset to introduce a classification workflow. Participants prepared a dataset, trained a model in the visual Azure Machine Learning Studio available at the time, and compared the result with the Kaggle task.

The original Slideshare player did not survive the site migration. Its reference was 155030146.

What the session was designed to teach

The visual workflow reduced the amount of code required for the demonstration. That let the session focus on the shape of a supervised-learning problem:

1. Define the outcome to predict.
2. Select training fields and handle missing values.
3. Split data for training and evaluation.
4. Train a classifier.
5. Review the result before treating it as useful.

The session did not show that visual tooling removes the need for data review, evaluation, or production controls.

Product-history note

Azure Machine Learning has changed since this talk. Interfaces, deployment choices, and product names shown in 2019 may no longer match the current service. Use current Microsoft Learn guidance before recreating the exercise.

The nine-part series

This was part 2 of a planned nine-session path covering Databricks, Spark ML, AutoML, MLflow, ML.NET, and production practices. Browse the Data Science for Dummies archive for the sessions that remain available.