Dave Kleinschmidt
Dave Kleinschmidt
This is a pretty substantial change in how non-special function calls are represented. Instead of generating an anonymous function, the new `FunctionTerm` simple wraps the called function, it's arguments (wrapped...
To be considered for the next breaking release (as discussed briefly in #172): - [ ] use keyword arguments for contructors - [ ] error on missing labels for hypothesis/contrasts...
This uses `Tables.Columns` as a (potentially) lightweight wrapper around input tables that does not convert them to the strongly typed `NamedTuple` of `Vector`s representation. This might make some things easier...
based on #174 (upstream user/repo seems to have been deleted) with a few tweaks
https://github.com/beacon-biosignals/Effects.jl/pull/23/ This comes up when you've already called `collect_matrix_terms` on a collection of terms that has a mix of matrix- and non-matrix-terms (e.g., a MixedEffects.jl RanefTerm), and then call `collect_matrix_terms`...
This is a meta-issue to keep track of major performance problems (broadly construed) that may require API changes and hence are blocking a 1.0 release. - [ ] type parameters...
We provide methods for many functions from StatsBase (like `coef`, `fit`, `predict`, `modelmatrix`, etc.) but don't export most of them. Most of the time this is fine as users will...
The current major blocker for getting rid of `TableStatisticalModel` wrappers (#32) and the `ModelFrame`/`ModelMatrix` structs is that they also keep a mask for the rows that are actually included from...
This is a "meta issue" to keep track of features that would be "nice to have" for contrast coding. I'll update this as other things come up. # Additional contrast...
There are some situations where you want to transform terms _before_ the schema is available, in part because this may affect how the schema itself is computed. For instance, for...