rådata

Rådata, short for rådata in Swedish, refers to data in its original, unprocessed form as collected from a source. Also called raw data, it has not been cleaned, transformed, or aggregated, and it often contains errors, missing values, duplicates, or inconsistencies. Rådata serves as the baseline for subsequent analysis and processing.

Sources of rådata include measurement instruments and sensors, logs from information systems, transaction records, surveys, and observational notes. It can be structured (tables), semi-structured (JSON, XML), or unstructured (text, images, audio). Raw data typically carries metadata such as timestamps, identifiers, units, and provenance, which are essential for interpretation and auditability.

The role of rådata in the data lifecycle is to provide an immutable reference point for analyses. Before analysis, raw data undergoes data preparation or preprocessing, which may involve quality checks, deduplication, handling missing values, normalization, transformation, and feature extraction. Maintaining data lineage and versioning is important to ensure reproducibility and accountability.

Formats and storage for raw data vary widely and include CSV, JSON, Parquet, log formats, and multimedia files. Rådata is often stored in data lakes or raw zones within data warehouses, with proper governance and access controls. Retention policies for raw data differ by domain, balancing reproducibility with privacy and storage costs.

Common challenges in handling rådata include noise, sensor drift, time synchronization issues, and biases introduced during collection. Privacy, security, and compliance considerations, such as anonymization and minimization, are also central when raw data contains potentially sensitive information.