Check data meaning, summarize variation and evaluate on held-out examples.
Free · No sign-in required3 lessonsWorked examples & practiceBeginner
0 of 3 lessons checked · Progress is saved in this browser.
01 / SYSTEM VIEWData quality
02 / SYSTEM VIEWAnalysis
03 / SYSTEM VIEWEvaluation
This map connects the course ideas; it is not a detailed engineering schematic.
LESSON 1 / 3 · FREE
Rows, units and missing data
Your goal: Identify what one record represents.
Understand the idea
Before calculating, define the observation, units and collection process. Missing is not the same as zero. Duplicates can overweight events, and unit mismatches can create convincing but meaningless averages. Preserve the original data and document cleaning decisions.
Worked example
A file contains 100 rows, including 8 duplicates and 2 additional unusable records. If those sets do not overlap, 90 usable unique records remain.
Try it yourself
Start with 120 rows and remove 10 duplicates plus 5 other unusable rows, without overlap. How many remain?
Your goal: Calculate a mean while noticing what it hides.
Understand the idea
The arithmetic mean adds values and divides by count. Its usefulness depends on comparable observations and the shape of the data. Outliers, changing operating regimes and unequal time intervals can make a simple mean misleading. Keep the distribution and context visible.
Worked example
Three equally weighted observations of 10, 12 and 20 have mean 14. The value 20 pulls the mean above the middle observation of 12.
Your goal: Keep evaluation information out of model fitting.
Understand the idea
A holdout represents unseen observations. Fit transformations using training data and apply them to evaluation data, rather than allowing evaluation information to shape the model. Time-series splits must respect time and the intended prediction task. An error score is meaningful only with a clear target.
Worked example
Actual values 10 and 20 with predictions 12 and 17 have absolute errors 2 and 3, so mean absolute error is 2.5 in the target units.
Try it yourself
Absolute prediction errors are 1, 2 and 6. Find their mean.
Explain each worked example in your own words, check the units and identify one assumption that could change the result. Then use the linked labs or roadmap to apply the ideas to a broader task.
Practice progress is a study aid, not a professional qualification or proof of field competence.
Continue in AI and Data Science
Apply these foundations in an available course, lab or roadmap. The linked resource’s access rules still apply.
These original lessons use simplified models and explicitly supplied practice values. Calculations teach concepts; they do not select real equipment, establish code compliance or authorize field work. Follow the linked references and advanced resources for deeper study.
Updated · Study progress stays in this browser. There is no certificate or paid gate for these lessons.