NA

Potential future improvements (TO DO LIST)

Converge improvements

• Offer alternative centre points for converge, such as st_point_in_polygon (method=c("centroid","pip")).
• Improve documentation, especially around the transformation of source spatial features which are in latlong projection in converge.
• Provide alternative transformation of source features in latlong projection? (e.g. alternative target crs?)

Manipulation improvements

• Offer to normalize dimensions of features, with st_normalize, so that full features fit within space of 0,1. (In this case, dimensions are lost, but shape is preserved.)
• Implement functions: normalize, scale.
• Consider renaming scale to avoid clash with base::scale, e.g. scaleByArea

Distribute improvements

• Additional distribute algorithms which translate from converged to distributed tableaus in different ways:-
  • regulargrid - the current algorithm, uses just the largest bounding box to grid a grid square
  • circlepack - using circle packing algorithm to pack differently sized polygons together in a less regular way
  • regularstacks - rather than filling rows or columns, this method would stack according to certain classificatory criteria, e.g. classifying parameters such as polygon area. (group.by="AREA", margin=1). Could be used as basis for spatial-icon-based “bar charts”, with features optionally normalized.
  • nofitpoly - using methods used to enable efficient cutting out irregular polygons from materials such as cloth or metal, this method would use a user-defined buffer around polygons to find heuristic-optimal translation matrices to fit all (buffered) polygons together, approximating a tesselation. (buffer=1.5).

Generalise transitions so that different sf objects can be matched based on a single feature.

• Add a z or byother parameter to functions so that objects can be transformed in a way dictated by another converged sf layer. This could potentially be used for rasters etc with some additional work
  • #’ @param z/byother An sf object, already processed with converge, from which transformation will be calculated. default=NULL.

Convenience functions

• Convenience textLabel function (with position options of centroid, within, above, below, left, right)
• Convenience scaleKey function (with option of linear vs. areal scales: value specification of linear and or density, as appropriate)
• Convenience functions for processing and plotting as one-liner (e.g. plot.sc.regulargrid, plot.sc.spatialbarplot)
• Implement a convenience function sc which performs the most likely chain with defaults, or a few minor options: converge -> distribute, so that it is possible to just call plot(sc(sf_layer)).

Other

• Integration with cartogram package?
• Optimisation of code (e.g. replacement of for loops on sf objects with faster alternatives?); low priority as it’s actually fast enough?