3 Data Cleaning and Preparation Free Online FlashCards

What is the goal of data cleaning?

Data cleaning addresses missing, duplicated, inconsistent, implausible, or unsuitable data formats so data is reliable for its intended use without losing its meaning.

How should conflicting duplicate records be resolved?

Use a defined rule, such as retaining the most recent verified record, rather than deleting conflicting records arbitrarily.

What should you preserve before cleaning data?

Keep an unchanged copy of the original data before making transformations. This preserves a reference for review and comparison.

What does a missing value tell you—and what does it not tell you?

A missing value means no usable value is recorded, but it does not explain why; it could be uncollected, inapplicable, unknown, or a failed measurement.

What should you assess about missing values before choosing a treatment?

Measure the number of missing values and check whether they are concentrated in particular columns or groups.

When might median imputation be appropriate, and how should imputed values be described?

Impute only when a justified estimate is needed; for a skewed numeric variable, the median may be appropriate. Record imputed values as estimates, not observations.

Why should missing numbers not automatically be replaced with zero?

Zero is a real value, so replacing missing numbers with zero can change summaries and misrepresent the data.

What should you check before removing a duplicate row?

Determine what one row represents first: a repeated row may be an accidental duplicate or a valid repeated event.

When is it appropriate to map inconsistent category labels together?

Map categories only after confirming that the different values are equivalent using a reliable definition; leave uncertain categories unknown or missing.

How should ordered categories be arranged?

Represent ordered categories, such as low, medium, and high, in their meaningful order rather than alphabetical order.

What is an outlier?

An outlier is an observation unusually far from the rest of a dataset.

How should an unusual value be handled?

Check unusual values against domain rules, source records, and related fields. Correct them only with supporting evidence; otherwise retain, flag, or use a suitable analysis method.