The flip Project
Jump to navigation
Jump to search
The flip Project is an open source R project. It consists of a series of R packages, beginning with the prefix of rhtml where they are HMTLwidgets, and flip otherwise. The word flip has no relevant meaning (it is not an acronym).
Objectives of the flip Project
The broad objective of the flip Project is to create a set of R packages that removes unnecessary complication from data analysis. In practice, this means that flip Project functions:
- Have telegraphic and verbose function and parameter names. For example, Regression rather than glm, and type = Binary Logit rather than family = binomial.
- Consistent treatment of missing values (e.g., building imputation and multiple imputation into multivariate methods as arguments).
- Support for lengthy labels containing punctuation in outputs.
- Outputs designed for easy and accurate interpretation.
- Support for sampling weights as the default method for dealing with weights.
- Support for subset where appropriate.
- Full-sample variable saving. For example, if the user uses predict.Regression, the resulting variable will be a vector with the same number of observations in the entire input data.frame. That is, it includes:
- Observations where the respondent had missing values on predictor variables (with a value of NA).
- Predicted values for observations with a missing value on the outcome variable but complete data for the missing values.
- Predicted values for observations excluded from the subset or due to missing weights information.
- Sensible defaults. Best likely choice, not simplest choice.
- Informative warnings that describe what the user should do to resolve the warnings.
Sacrifices
These objectives are not without costs. In particular, functions written for the flip Project are generally:
- Verbose.
- Complicated.
- Memory-intensive.