- Notifications
You must be signed in to change notification settings - Fork3
echasnovski/comperes
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Repository files navigation
comperes offers a pipe (%>%) friendly set of tools for storing andmanaging competition results. Understanding ofcompetition is quitegeneral: it is a set ofgames (abstract event) in whichplayers(abstract entity) gain some abstractscores (typically numeric). Themost natural example is sport results, however not the only one. Forexample, product rating can be considered as a competition betweenproducts as “players”. Here a “game” is a customer that reviews a set ofproducts by rating them with numerical “score” (stars, points, etc.).
This package leveragesdplyr’s grammar ofdata manipulation. Only basic knowledge is enough to usecomperes.
comperes provides the following functionality:
- Store and convert competition results:
- Inlong format as atibble withone row per game-player pair. Functions:
as_longcr(),is_longcr(). - Inwide format as a
tibblewith one row per game with fixedamount of players. Functions:as_widecr(),is_widecr().
- Inlong format as atibble withone row per game-player pair. Functions:
- Summarise:
- Computeitem summaries with functions using
dplyr’s grammar.Functions:summarise_item(),summarise_game(),summarise_player(). - Compute andjoin item summaries to data for easy transformation.Functions:
join_item_summary(),join_game_summary(),join_player_summary().
- Computeitem summaries with functions using
- Compute Head-to-Head values (a summary statistic of directconfrontation between two players) with functions also using
dplyr’sgrammar:- Store output inlong format as a
tibblewith one row per pair ofplayers. Function:h2h_long(). - Store output inmatrix format as a matrix with rows and columnsdescribing players and entries - Head-to-Head values. Function:
h2h_mat(). - Usecommon Head-to-Head functions with rlang’s unquotingmechanism. Example:
. %>% h2h_mat(!!!h2h_funs["num_wins"]).
- Store output inlong format as a
You can installcomperes from CRAN with:
install.packages("comperes")To install the most recent development version from GitHub use:
# install.packages("devtools")devtools::install_github("echasnovski/comperes")
We will be usingncaa2005, data fromcomperes package. It is anexample competition results (hereafter - results) of an isolated groupof Atlantic Coast Conference teams provided in book“Who’s#1”by Langville and Meyer. It looks like this:
library(comperes)ncaa2005#> # A longcr object:#> # A tibble: 20 × 3#> game player score#> <int> <chr> <int>#> 1 1 Duke 7#> 2 1 Miami 52#> 3 2 Duke 21#> 4 2 UNC 24#> 5 3 Duke 7#> 6 3 UVA 38#> # … with 14 more rows
This is an object of classlongcr which describes results in long form(each row represents the score of particular player in particulargame). Because in this competition a game always consists from twoplayers, more natural way to look atncaa2005 is in wide format:
as_widecr(ncaa2005)#> # A widecr object:#> # A tibble: 10 × 5#> game player1 score1 player2 score2#> <int> <chr> <int> <chr> <int>#> 1 1 Duke 7 Miami 52#> 2 2 Duke 21 UNC 24#> 3 3 Duke 7 UVA 38#> 4 4 Duke 0 VT 45#> 5 5 Miami 34 UNC 16#> 6 6 Miami 25 UVA 17#> # … with 4 more rows
This convertedncaa2005 into an object ofwidecr class whichdescribes results in wide format (each row represents scores of allplayers in particular game). Allcomperes functions expect either adata frame with results structured in long format or one of supportedclasses:longcr,widecr.
Withcompere the following summaries are possible:
ncaa2005 %>% summarise_player(min_score= min(score),mean_score= mean(score))#> # A tibble: 5 × 3#> player min_score mean_score#> <chr> <int> <dbl>#> 1 Duke 0 8.75#> 2 Miami 25 34.5#> 3 UNC 3 12.5#> 4 UVA 5 18.5#> 5 VT 7 33.5# Using list of common summary functionslibrary(rlang)ncaa2005 %>% summarise_game(!!!summary_funs[c("sum_score","num_players")])#> # A tibble: 10 × 3#> game sum_score num_players#> <int> <int> <int>#> 1 1 59 2#> 2 2 45 2#> 3 3 45 2#> 4 4 45 2#> 5 5 50 2#> 6 6 42 2#> # … with 4 more rows
Supplied list of common summary functions has 8 entries, which arequoted expressions to be used indplyr grammar:
summary_funs#> $min_score#> min(score)#>#> $max_score#> max(score)#>#> $mean_score#> mean(score)#>#> $median_score#> median(score)#>#> $sd_score#> sd(score)#>#> $sum_score#> sum(score)#>#> $num_games#> length(unique(game))#>#> $num_players#> length(unique(player))ncaa2005 %>% summarise_player(!!!summary_funs)#> # A tibble: 5 × 9#> player min_score max_score mean_score median_score sd_score sum_score num_games num_players#> <chr> <int> <int> <dbl> <dbl> <dbl> <int> <int> <int>#> 1 Duke 0 21 8.75 7 8.81 35 4 1#> 2 Miami 25 52 34.5 30.5 12.3 138 4 1#> 3 UNC 3 24 12.5 11.5 9.40 50 4 1#> 4 UVA 5 38 18.5 15.5 14.0 74 4 1#> 5 VT 7 52 33.5 37.5 19.9 134 4 1
To modify scores based on the rest of results one can usejoin_*_summary() functions:
suppressPackageStartupMessages(library(dplyr))ncaa2005_mod<-ncaa2005 %>% join_player_summary(player_mean_score= mean(score)) %>% join_game_summary(game_mean_score= mean(score)) %>% mutate(score=player_mean_score-game_mean_score)ncaa2005_mod#> # A longcr object:#> # A tibble: 20 × 5#> game player score player_mean_score game_mean_score#> <int> <chr> <dbl> <dbl> <dbl>#> 1 1 Duke -20.8 8.75 29.5#> 2 1 Miami 5 34.5 29.5#> 3 2 Duke -13.8 8.75 22.5#> 4 2 UNC -10 12.5 22.5#> 5 3 Duke -13.8 8.75 22.5#> 6 3 UVA -4 18.5 22.5#> # … with 14 more rowsncaa2005_mod %>% summarise_player(mean_score= mean(score))#> # A tibble: 5 × 2#> player mean_score#> <chr> <dbl>#> 1 Duke -15.5#> 2 Miami 11.4#> 3 UNC -5#> 4 UVA -2.12#> 5 VT 11.2
This code modifiesscore to be average player score minus average gamescore. Negative values indicate poor game performance.
Computation of Head-to-Head performance is done withh2h_long()(output is a tibble; allows multiple Head-to-Head values per pair ofplayers) orh2h_mat() (output is a matrix; only one value per pair ofplayers).
Head-to-Head functions should be supplied indplyr grammar but forplayers’ matchups: direct confrontation betweenordered pairs ofplayers (including playing with themselves) stored in wide format:
ncaa2005 %>% get_matchups()#> # A widecr object:#> # A tibble: 40 × 5#> game player1 score1 player2 score2#> <int> <chr> <int> <chr> <int>#> 1 1 Duke 7 Duke 7#> 2 1 Duke 7 Miami 52#> 3 1 Miami 52 Duke 7#> 4 1 Miami 52 Miami 52#> 5 2 Duke 21 Duke 21#> 6 2 Duke 21 UNC 24#> # … with 34 more rows
Typical Head-to-Head computation is done like this:
ncaa2005 %>% h2h_long(mean_score_diff= mean(score1-score2),num_wins= sum(score1>score2) )#> # A long format of Head-to-Head values:#> # A tibble: 25 × 4#> player1 player2 mean_score_diff num_wins#> <chr> <chr> <dbl> <int>#> 1 Duke Duke 0 0#> 2 Duke Miami -45 0#> 3 Duke UNC -3 0#> 4 Duke UVA -31 0#> 5 Duke VT -45 0#> 6 Miami Duke 45 1#> # … with 19 more rowsncaa2005 %>% h2h_mat(mean(score1-score2))#> # A matrix format of Head-to-Head values:#> Duke Miami UNC UVA VT#> Duke 0 -45 -3 -31 -45#> Miami 45 0 18 8 20#> UNC 3 -18 0 2 -27#> UVA 31 -8 -2 0 -38#> VT 45 -20 27 38 0
Supplied list of common Head-to-Head functions has 9 entries, which arealso quoted expressions:
h2h_funs#> $mean_score_diff#> mean(score1 - score2)#>#> $mean_score_diff_pos#> max(mean(score1 - score2), 0)#>#> $mean_score#> mean(score1)#>#> $sum_score_diff#> sum(score1 - score2)#>#> $sum_score_diff_pos#> max(sum(score1 - score2), 0)#>#> $sum_score#> sum(score1)#>#> $num_wins#> num_wins(score1, score2, half_for_draw = FALSE)#>#> $num_wins2#> num_wins(score1, score2, half_for_draw = TRUE)#>#> $num#> dplyr::n()ncaa2005 %>% h2h_long(!!!h2h_funs)#> # A long format of Head-to-Head values:#> # A tibble: 25 × 11#> player1 player2 mean_score_diff mean_…¹ mean_…² sum_s…³ sum_s…⁴ sum_s…⁵ num_w…⁶ num_w…⁷ num#> <chr> <chr> <dbl> <dbl> <dbl> <int> <dbl> <int> <dbl> <dbl> <int>#> 1 Duke Duke 0 0 8.75 0 0 35 0 2 4#> 2 Duke Miami -45 0 7 -45 0 7 0 0 1#> 3 Duke UNC -3 0 21 -3 0 21 0 0 1#> 4 Duke UVA -31 0 7 -31 0 7 0 0 1#> 5 Duke VT -45 0 0 -45 0 0 0 0 1#> 6 Miami Duke 45 45 52 45 45 52 1 1 1#> # … with 19 more rows, and abbreviated variable names ¹mean_score_diff_pos, ²mean_score,#> # ³sum_score_diff, ⁴sum_score_diff_pos, ⁵sum_score, ⁶num_wins, ⁷num_wins2
To compute Head-to-Head for only subset of players or include values forplayers that are not in the results, use factorplayer column:
ncaa2005 %>% mutate(player=factor(player,levels= c("Duke","Miami","Extra"))) %>% h2h_mat(!!!h2h_funs["num_wins"],fill=0)#> # A matrix format of Head-to-Head values:#> Duke Miami Extra#> Duke 0 0 0#> Miami 1 0 0#> Extra 0 0 0
About
Manage Competition Results
Resources
License
Uh oh!
There was an error while loading.Please reload this page.
Stars
Watchers
Forks
Packages0
Uh oh!
There was an error while loading.Please reload this page.