saveRDSreadRDS

saveRDS and readRDS are base R functions designed to serialize a single R object to disk and to restore it later. They provide a compact, language-native mechanism for persisting objects, sharing them between sessions, or caching results. Unlike the save function, which writes one or more named objects to an .RData file, saveRDS and readRDS focus on a single object and do not save the object’s original name.

saveRDS writes a single R object to a file in the RDS format. Its typical usage is saveRDS(object, file) with optional arguments such as ascii, version, compress, and refhook. The ascii option writes a human-readable representation when supported, version selects the serialization format version, compress controls whether compression is used, and refhook allows handling of external pointers during restoration. Importantly, saveRDS does not store the object name; the name is assigned when the object is read back.

readRDS reads an RDS file and returns the stored object. The returned value is typically assigned to a variable, e.g., obj <- readRDS("path/model.rds"). The original object name is not restored, so explicit assignment determines how the object is named in the current session. readRDS accepts a refhook argument similar to saveRDS for advanced restoration scenarios.

Compared with save and load, saveRDS/readRDS are ideal for persisting single objects and controlling their loaded names, while save/load handle multiple objects and preserve their names within an environment. The .rds format is R-specific and primarily used for R-to-R data exchange.