4 Data Cleaning and Preparation Online Quiz Questions

True

False

Treat zero, an empty string, and “none” as missing in every column.

Identify and standardize missing-value markers, while distinguishing them from zero and other actual values.

Count only blank cells as missing, because other markers are valid values.

Replace every missing value with zero before counting.

Parse it as March 4 because month-first order is universal.

Parse it as April 3 because day-first order is universal.

Confirm whether the source uses month-first or day-first order before parsing.

Replace it with a missing value because it is not already in ISO format.

Check that dates parse as dates and quantities are numeric.

Check that required identifiers are present.

Check that a status belongs to its defined set of allowed values.

Check that a delivery date does not precede its order date.

Check that fields defined as unique contain no duplicates.

Treat every negative amount as invalid without investigating its meaning.

Require every column to have a value in every row.

True

False

Split the data into training and test sets before learning imputation values.

Fit the imputation transformation using training data only.

Apply the fitted training transformation to the test data.

Calculate an imputation value using the full dataset before splitting.

Fit a separate imputation transformation on the test set.

Convert every value using the same numerical multiplier regardless of its original unit.

Verify the original and target units before converting.

Treat unit labels as cosmetic and remove them before analysis.

Convert every measurement to inches because all datasets use that standard.

Automatically replace in progress with open because it sounds similar.

Delete the row because any unlisted status proves the record is false.

Flag the value for investigation against the defined set of allowed statuses.

Expand the allowed set to include every value found, without review.

True

False

Replace unusual values directly in the only copy of the data.

Preserve the original data and make justified changes to a working copy.

Delete every row containing an unexpected value.

Standardize all values before checking what the columns mean.

Treat zero, an empty string, and “none” as missing in every column.

Count only blank cells as missing and leave other markers unchanged.

Identify and standardize the column’s missing-value markers before counting missing values.

Replace every missing value with zero before checking how many there are.

Convert every measurement to the unit used most often in the dataset without checking its source.

Confirm the source units and the intended meaning before converting measurements.

Treat values in different units as equivalent if their column names match.

Remove measurements whenever the source unit is not already standardized.

Delete one record whenever two rows have the same date.

Treat matching dates and amounts as proof that the records are duplicates.

Keep all repeated rows without checking what makes a record unique.

Define a meaningful record key, then inspect suspected duplicates before removing any.