📄️ Streamlit
Streamlit for data scientists: turn a trained churn model into a shareable web app, from a first form to a deployment protected by a password.
📄️ 1. Execution model
Module 1 of the Streamlit premium course: install the framework, run your first script, and understand why the whole file is re-executed on every interaction.
📄️ 2. Components
Module 2 of the Streamlit premium course: text inputs, sliders, selectors, dates, metrics and tables, combined into a real scoring form for one customer.
📄️ 3. Layout
Module 3 of the Streamlit premium course: organize a page with columns, tabs, expanders and a sidebar for filters, without CSS and without a front-end framework.
📄️ 4. Charts and tables
Module 4 of the Streamlit premium course: native charts, Plotly and Altair figures, interactive and editable data grids, applied to the churn dashboard.
📄️ 5. Caching
Module 5 of the Streamlit premium course: cache_data versus cache_resource, when to invalidate, and the traps of mutable objects and unhashable arguments.
📄️ 6. State and forms
Module 6 of the Streamlit premium course: persist values across reruns with session_state, batch inputs inside st.form, and drive multi-step navigation.
📄️ 7. Upload and download
Module 7 of the Streamlit premium course: upload a CSV of customers, validate its columns, score the file in batch, and offer the results as a download.
📄️ 8. Calling a model
Module 8 of the Streamlit premium course: call a local model or a remote inference API, with timeouts, retries, error handling and a loading indicator.
📄️ 9. Theme and pages
Module 9 of the Streamlit premium course: brand the app through the config file, add a logo and help messages, order actions, and split into a multipage application.
📄️ 10. Deployment
Module 10 of the Streamlit premium course: deploy on Community Cloud or a Docker container, keep secrets out of git, protect the app with a password or delegated authentication.
📄️ Recap and exam
Complete recap of the Streamlit premium course: execution model, widgets, layout, caching, state, uploads, model calls, theme and deployment, then the 40-question exam.