Build on your SQL & BI skills / CORE PATH
Optional · a clean ML bridge
Carry reliable Python into model preparation and evaluation.
Build on your ML direction after data contracts and reproducibility are solid.
Understand the idea
Fit preprocessing on training data and apply the learned transformation to held-out data. Splitting after imputing or scaling leaks information about the evaluation set.
Document the prediction target, observation time and availability of each feature. A feature that becomes known after the business decision can make an apparently strong model unusable.
What to learn
- Split data before fitting preprocessing.
- Keep transformations inside a pipeline.
- Choose an honest baseline and a metric aligned to the task.
Practice on your own data
Write a leakage checklist and a small baseline experiment on an appropriately licensed dataset.
Practice this lesson in the Python lab →
The held-out set never fits the imputer, encoder or scaler.
Prerequisites
Build 02 · PixelFed quality CLI · NumPy & time-aware analysis
Video and playlist companions
Choose one explanation for the concept, then attempt the practice. These are topic mappings, not a claim that every linked lesson covers every outcome.
03 تعليم الآلة , القسم الثالث : بايثون Machine learning , Python
Loads YouTube on request. Open on YouTube ↗ if the owner restricts embedding.
Data Exploration and Preparation
Loads YouTube on request. Open on YouTube ↗ if the owner restricts embedding.
Sources and further reading
- scikit-learn · getting started ↗Official documentation
- 03 تعليم الآلة , القسم الثالث : بايثون Machine learning , Python ↗User-selected YouTube playlist · also in the inherited DE catalog · Inherited from data-engineering-learning-roadmap: Ahmed Shaaban mentorship
- Data Exploration and Preparation ↗User-selected YouTube playlist
Track progress on this topic → · Markdown source ↗