Fig 2. A five-step data cleaning algorithm for growth data that uses pre-defined measurement predictions and prediction limits to identify which measurements are likely to be erroneous and to make appropriate corrections and deletions.