recoding

Recoding is the process of transforming information from one encoding, representation, or classification to another. It is used when data must be consumed by different systems, when analyses require a different level of granularity, or when cross-disciplinary comparison is needed. In some cases recoding is lossless; in others it can change semantics if the encodings differ.

In computing, recoding typically means converting text between character encodings (for example, ISO-8859-1 to UTF-8) or reformatting data. It also covers re-encoding signals or media formats. Proper recoding preserves data integrity; incorrect mappings can cause mojibake or data loss.

In statistics and data science, recoding refers to redefining variable values or categories, such as collapsing response options or creating binary indicators. It is a data-cleaning step that improves interpretability or model compatibility, and should be documented to enable replication.

In biology, translational recoding describes cases where the genetic code is read in an alternative way during protein synthesis, including programmed ribosomal frameshifting, stop-codon readthrough, and incorporation of nonstandard amino acids. Recoding allows greater coding capacity and regulatory control, notably in viruses and some organisms.

In linguistics and social science, recoding can mean reexpressing information in a different linguistic form or label set, such as paraphrasing, code-switching, or recoding survey categories. Clear rationale and documentation are essential to maintain transparency and comparability across studies.