Common cleaning tasks
Types, blanks, splitting and combining, filtering, renaming.
The transformations you will reach for most, and when each is the right tool.
Change a column's type
The most common fix by far. Text that should be a number cannot be summed; text that should be a date cannot be grouped by month or used in time intelligence.
Handle blanks
Decide explicitly what a blank means. Sometimes it is genuinely zero; sometimes it means "unknown", and turning it into zero quietly changes every average built on the column. Choosing deliberately is the point.
Split and combine columns
Split a full name into first and last, or a compound code into its parts. Combine columns when the useful unit is the pair rather than either half.
Filter rows
Remove rows that should never take part — test records, cancelled orders, a partial final month. Filtering here removes them from every chart at once, which is the difference between a rule and a habit of remembering.
Rename
Column names appear in field lists, chart legends and axis labels. cust_nm_1 works; Customer
name is what a reader should see. Renaming early means every chart inherits the readable name.
A note on order
Because steps are applied in sequence, put type corrections early — several other steps behave differently depending on whether they are working with text or with a real date or number.