Reshaping and aggregating data: an introduction to reshape package

Zhongheng Zhang


It is common that data format extracted from clinical database does not meet the purpose of statistical analysis. In clinical research, variables are frequently measured repeatedly over the follow-up period. Such data can be displayed either in wide or long format. Transformation between these 2 forms can be challenging by hand. Fortunately, there are sophisticated packages in R environment. Data frame should firstly be melted and then casted to format that you want. Aggregation over unique combination of id variables is also allowable. Additionally, the article also introduces 2 functions colsplit() and funstofun() that can be useful in some circumstances.