In znmeb/dfstools: Tidy Data Analytics from Sports Data APIs

Load the input data

library(readr)
library(dplyr)
library(tibble)
library(dfstools)
library(knitr)
RegularSeasonCompactResults <- read_csv(
  "~/DFS/kaggle/RegularSeasonCompactResults.csv",
  col_types = cols(
    DayNum = col_integer(), 
    LScore = col_integer(), 
    LTeamID = col_integer(),
    NumOT = col_integer(), 
    Season = col_integer(),
    WScore = col_integer(), 
    WTeamID = col_integer())) %>% 
  filter(Season == 2019)
SampleSubmissionStage2 <- read_csv("~/DFS/kaggle/SampleSubmissionStage2.csv")
Teams <- read_csv(
  "~/DFS/kaggle/Teams.csv",
  col_types = cols(
    FirstD1Season = col_integer(),
    LastD1Season = col_integer(), 
    TeamID = col_integer()))

Wrangle the data into an mvglmmRank game.data input file

game_data <- kaggle_game_data(RegularSeasonCompactResults)

Run the fast model (NB.mov)

nb_mov_model <- mvglmmRank_model(game_data)

Compute fast rating tibble

nb_mov_ratings <- kaggle_ratings(nb_mov_model, Teams)
nb_mov_ratings$TeamID <- as.character(nb_mov_ratings$TeamID)
nb_mov_ratings %>% top_n(16) %>% select(-TeamID) %>% kable(
  row.names = TRUE, digits = 2, format = "pandoc", 
  caption = "Men's NCAA® Ratings")

Fill out the submission

nb_mov_submission <- SampleSubmissionStage2
for (i in 1:nrow(nb_mov_submission)) {
  nb_mov_submission$Pred[i] <- kaggle_probability(
    nb_mov_model, nb_mov_submission$ID[i])
}
write_csv(
  nb_mov_submission, path = "~/DFS/kaggle/mens_2019_nb_mov_submission.csv")

Session info

sessionInfo()

znmeb/dfstools documentation built on March 3, 2020, 5:50 p.m.