Skip to content

IAI Actuarial Core Principles · Actuarial Statistics · Exploratory data analysis

A life insurer records the age at death of policyholders in a file. Ages are shown for 400 records, but 12 records show age 0 and 8 records show age 999. Which is the most appropriate first step in data cleaning?

Investigate the records before changing them. The value 999 is likely a missing-data code that should be recoded as missing, and age 0 may be a real infant death or a default entry. Blind deletion or mean replacement risks introducing bias.

  1. ADelete all 20 records at once because they are outliers
  2. BReplace all 20 values with the mean age of the remaining records
  3. CInvestigate the source: 999 is probably a missing-value code to be recoded, and age 0 needs checking as a genuine infant death or a default entryCorrect
  4. DTreat 999 as a genuine age and use a robust estimator
  5. Ignore the issue since 20 of 400 is only 5%

Explanation

Values like 999 are typically placeholder codes for missing data and should be recoded, not analysed as ages. Zeros may be genuine or defaults, so they must be checked against source records before any deletion or imputation. Blind deletion or mean replacement can bias results.

Did you get it right without looking?

One question tells you little. A timed set on Exploratory data analysis shows your real accuracy, how long you take and where you lose marks.

Practise Exploratory data analysisFree diagnostic mock

More Exploratory data analysis questions