School Nutrition Programme Data Cleansing

Beneficiary data was collected from many schools using different reporting formats. After several years it contained duplicates and empty values, making beneficiary counts inconsistent between departments.

We found 126 duplicate national ID numbers across 68,367 rows, plus empty values in identity columns and inconsistent birth date formats. Some cases had the same ID number with different names, so they could not simply be merged.

We defined a staged cleansing process: inventory duplicates, normalise empty values, validate ID and date formats, then isolate cases needing manual tracing. Every rule was tested on sample data before being applied broadly.

Beneficiary counts are now consistent between departments. Validation rules were applied at data entry so the same problems do not recur in the next period.