The map-exts package

[maintain]

Please see README.md


[Skip to ReadMe]

Properties

Versions0.1.0.0, 0.1.0.0, 0.1.0.1
Dependenciesbase (>=4.7 && <5), bytestring, cassava, containers [details]
LicenseBSD3
Copyright2016 Elsen, Inc
AuthorCharles Cooper
Maintainercooper.charles.m@gmail.com
CategoryData Structures
Home pagehttp://github.com/elsen-trading/map-extensions#readme
Executablestitanic-example, example
UploadedThu Apr 14 21:45:51 UTC 2016 by coopercm

Modules

[Index]

Downloads

Maintainers' corner

For package maintainers and hackage trustees

Readme for map-exts-0.1.0.0

map-extensions

This module contains some extensions to Data.Map. Some of them are convenience functions.

It also contains functions to support a split-apply-combine workflow, by representing labeled, multi-dimensional data as multiply nested Maps. For instance, a two dimensional matrix with one axis indexed by 'Name's and the other axis labeled by 'Job's, we would represent such a structure with a Lookup2 Name Job Double. Such a structure is not terribly efficient (it takes O(n log(n)) space and O(log(n)) time for insert/update/delete operations with high constant factor owing to all the pointer manipulation), but it is expressive.

split : groupMapBy apply : fmap combine : foldr/foldMap reshape : transpose

Say we had a dataFrame :: (Lookup2 Name Job Age). We could group by Job by running transpose on it. If we wanted to find the average Age in a job, we could define

mean :: Map k v -> Int
mean xs = foldr (+) 0 xs / Map.size xs

type Name = String
type Job  = String
type Age  = Double
averageAgeInJob :: Lookup2 Name Job Age -> Map Job Age
averageAgeInJob dataFrame = fmap mean $ transpose dataFrame

This illustrates applying an aggregation over a particular index by reshaping with transpose, defining a fold, and then focusing it inside the structure using fmap.