Скачать или смотреть Day 3 Lecture — Feature Engineering for Household Power Data ⚡ | ML Summer School 2025

Set your models up for success by turning raw power-consumption logs into machine-learning gold!
This is Day 3 of our free 15-Day AI & Machine Learning Summer School (July 7 – 28 2025).

⏱️ Layout
Intro & why feature engineering matters
Quick recap of the dataset (voltage, active/reactive power, sub-meters)
Parsing the `datetime` column → time-aware objects
Selecting key numeric columns (power, voltage, intensity, sub-meters)
Ensuring data-type integrity (`astype(float)`)
Extracting hour of day ➜ capturing diurnal patterns
Extracting day of week ➜ weekday vs weekend behavior
Creating Apparent Power (active + reactive) — formula walk-through
Scaling data — `StandardScaler` vs `MinMaxScaler`
Saving the engineered dataset to *CSV* for reuse
Best practices & Day 3 recap
Sneak peek: Day 4 (Baseline ML Models)

#📚 Resources

• PPT Slide and Code – see GitHub repo
• Jupyter notebook with full code – see GitHub repo
• Course GitHub repo – https://github.com/graceugochinneji/C...

#### 🔗 Helpful Docs

Pandas – [https://pandas.pydata.org/docs/](https://pandas.pydata.org/docs/)
Scikit-learn (`StandardScaler`, `MinMaxScaler`) – [https://scikit-learn.org/stable/](https://scikit-learn.org/stable/)
Python datetime – [https://docs.python.org/3/library/dat...](https://docs.python.org/3/library/dat...)

✨ What you’ll learn

The why and how of feature engineering for time-series energy data
Crafting temporal features (hour, weekday) to capture human power-usage habits
Deriving Apparent Power to enrich model signals
Choosing and applying the right scaling technique
Saving a clean, reproducible, model-ready dataset

👍 Like this video if it helps, comment your feature-engineering questions, and subscribe & 🔔 to stay on track for Day 4!

AI & Machine Learning Summer School is your 15-day, code-first journey from beginner to AI builder. Hosted by CDUT, we release a new lesson every day—complete with slides, notebooks, and real-world datasets. Topics include #environment setup, #EDA, #feature engineering, classical ML, #deep learning, #MLOps, and #interactive dashboards. Join thousands of learners leveling-up their AI skills—subscribe and start coding with us today!