Data preparation

Data preparation converts raw observations into a form suitable for analysis while preserving their scientific meaning.

Typical checks

Important steps include checking units, variable definitions, dates and time intervals, missing values, duplicated records, impossible values and changes in measurement procedures.

Transformations

Some analyses require scaling, normalisation, aggregation or transformation. Each transformation should have a clear purpose because it changes the quantities being fitted or interpreted.

Reproducibility

Raw data should normally be retained separately, and preparation steps should be recorded in code or a documented workflow.