Airball_Download.Rmd

---
title: "Airball NBA Download"
author: Jose Fernandez
2021
---


```{r}
library(tidyverse)
library(airball) #more info: https://github.com/josedv82/airball

#data pull using nba_travel() function from {airball} package.
data <- nba_travel(start_season = 2000, end_season = 2019) %>% ungroup() # adjust parameters based on interest

#feature engineering
dat <- data %>% 
  group_by(Season, Team) %>%
  
  #add and clean missing dates needed for time between games
  complete(Date = seq.Date(min(Date), max(Date), by="day")) %>% 
  mutate(Phase = ifelse(is.na(Phase), "-", Phase)) %>%
  
  #calcuate number of games playe in different time windows
  mutate(G3 = zoo::rollapplyr(Phase == "RS", width = 3, sum, partial = T)) %>%
  mutate(G5 = zoo::rollapplyr(Phase == "RS", width = 5, sum, partial = T)) %>%
  mutate(G7 = zoo::rollapplyr(Phase == "RS", width = 7, sum, partial = T)) %>%
  mutate(G9 = zoo::rollapplyr(Phase == "RS", width = 9, sum, partial = T)) %>%
  mutate(G11 = zoo::rollapplyr(Phase == "RS", width = 11, sum, partial = T)) %>%
  mutate(G13 = zoo::rollapplyr(Phase == "RS", width = 13, sum, partial = T)) %>%
  mutate(G15 = zoo::rollapplyr(Phase == "RS", width = 15, sum, partial = T)) %>%
  mutate(G17 = zoo::rollapplyr(Phase == "RS", width = 17, sum, partial = T)) %>%
  mutate(G19 = zoo::rollapplyr(Phase == "RS", width = 19, sum, partial = T)) %>%
  mutate(G21 = zoo::rollapplyr(Phase == "RS", width = 21, sum, partial = T)) %>%
  
  #calculate number of time zone changes for different time windows
  mutate(`Shift` = ifelse(is.na(`Shift (hrs)`), 0, `Shift (hrs)`)) %>%
  mutate(Shift3 = zoo::rollapplyr(Shift != 0, width = 3, sum, partial = T)) %>%
  mutate(Shift5 = zoo::rollapplyr(Shift != 0, width = 5, sum, partial = T)) %>%
  mutate(Shift7 = zoo::rollapplyr(Shift != 0, width = 7, sum, partial = T)) %>%
  mutate(Shift9 = zoo::rollapplyr(Shift != 0, width = 9, sum, partial = T)) %>%
  
  #calculate distance traveled for different time windows
  mutate(Distance = ifelse(is.na(Distance), 0, Distance)) %>%
  mutate(Distance3 = zoo::rollapplyr(Distance, width = 3, sum, partial = T)) %>%
  mutate(Distance5 = zoo::rollapplyr(Distance, width = 5, sum, partial = T)) %>%
  mutate(Distance7 = zoo::rollapplyr(Distance, width = 7, sum, partial = T)) %>%
  mutate(Distance9 = zoo::rollapplyr(Distance, width = 9, sum, partial = T)) %>%
  filter(Phase != "-") %>%
  
  #calculate winning %
  mutate(Games = row_number()) %>%
  mutate(W = zoo::rollapplyr(`W/L` == "W", 82, sum, partial = T)) %>%
  mutate(win_pct = round((W/Games),2)) %>% 
  
  #calculate winnning streak
  mutate(Streak2 = sequence(rle(`W/L`)$lengths)) %>% 
  mutate(Streak2 = ifelse(`W/L` == "L", Streak2 * -1, Streak2)) %>%
  mutate(Streak1 = lag(Streak2)) %>%
  mutate(Streak = ifelse(is.na(Streak1), 0, Streak1)) %>%
  select(-Streak2, -Streak1) %>%
  ungroup() %>% 
  
  #select features of interest
  select(Season, Team, Opponent, Date, Location, Distance, Distance3, Distance5, Distance7, Distance9, Rest, Zone = `Shift (hrs)`, Result = `W/L`, G3, G5, G7, G9, G11, G13, G15, G17, G19, G21, Shift3, Shift5, Shift7, Shift9, Streak, win_pct) %>% 
  
  #final cleaning
  mutate(Location = ifelse(Location == "Away", 1, 0)) %>% 
  mutate(Result = ifelse(Result == "W", 1, 0)) %>%
  mutate(Result = ifelse(Date >= Sys.Date(), NA, Result)) %>%
  mutate(win_pct = ifelse(Date >= Sys.Date(), NA, win_pct)) %>%
  filter(Season == "2020-21") %>% #comment this for all data
  fill(win_pct)

#code to get the same features for the opponent team so that in can be use for analysis as well
dat1 <- dat %>%
  select(Season, Opponent = Team, Team = Opponent, Date, Opp_Distance = Distance, Opp_Rest = Rest, Opp_Zone = Zone, Opp_G3 = G3, Opp_G5 = G5, Opp_G7 = G7, Opp_G9 = G9, Opp_G11 = G11, Opp_G13 = G13, Opp_G15 = G15, Opp_G17 = G17, Opp_G19 = G19, Opp_G21 = G21, Opp_Shift3 = Shift3,  Opp_Shift5 = Shift5,  Opp_Shift7 = Shift7, Opp_Shift9 = Shift9, Opp_Distance3 = Distance3, Opp_Distance5 = Distance5, Opp_Distance7 = Distance7, Opp_Distance9 = Distance9, Opp_Streak = Streak, opp_win_pct = win_pct)

#join team and opponents datasets
dat_full <- full_join(dat, dat1) 

#uncomment to export the data into csv format
#write.csv(dat_full, "data_NBA_sche.csv")

```