BaseballSimulations

##  Generating Batters/Pitchers Tutorial
##  Evan Boyd
##  enboyd@wisc.edu
##  @eboyd42

#Welcome to R! I will try and go over all of what I did through the comments (using #). 
#If you have any questions about what I did, you can always message me on Twitter or email me.
#Google/Stack Overflow is also a great resource.

##I have had some trouble reading files off of websites lately, but luckily,
##you can download a CSV file from baseball reference and import it straight in!
##Links to the tables I used:
#Batter: https://www.baseball-reference.com/leagues/MLB/bat.shtml
#Pitcher: https://www.baseball-reference.com/leagues/MLB/pitch.shtml
#Save the totals table as a CSV to somewhere on your computer.
setwd("C:/Users/Evan Boyd/Desktop/R Work/Random Simulations") 

batters <- read.csv("BatterTotals.csv", header = T)
pitchers <- read.csv("PitcherTotals.csv", header = T)

head(batters)
colnames(batters)
#You can look at just a specific column by using the $ command after the title of your table.
batters$HR

batters$HR[1]/batters$AB[1] #Find percentage of home runs in 2017 season, using total # of HR / # of AB
batters[1,13]/batters[1,8] #Row 1, column corresponding to HR/AB, respectively (same as above command)

#I will find all of the percentages using a loop. We know our column corresponding to "AB" is column 8.
#Now all we have to do is find the percentages for HR, walk, SO, 1B, 2B, and 3B.
#I will put this in a data.frame (a table), so I can see them all at once.
#We already have two data frames - one for batters, and one for hitters. We can find this using the class() command.
class(batters)
class(hitters)
#Note that data frames can contain strings, integer values, and character values in different columns.
class(batters$Year)

#To get singles, I need to subtract hits by 2B, 3B, and HR. I can add that as a column to my batters data frame.
batters$Singles = batters$H - batters$X2B - batters$X3B

#Let's look only at the 2017 data. I can switch the batters data frame so it is only looking at the first row, 2017.
batters = batters[1,]

percentages <- data.frame(Singles = 0, Doubles = 0, Triples = 0, HR = 0, Walk = 0, SO = 0) #Creating a column for each stat.
#I will now assign values to each column in the data frame. I could write a loop for this, but there are
#enough values where I can do this by hand. Also I'm lazy.

percentages$Singles = batters$Singles/batters$AB #It does not really matter if you use '=' or '<-'
percentages$Doubles = batters$X2B/batters$AB
percentages$Triples = batters$X3B/batters$AB
percentages$HR = batters$HR/batters$AB
percentages$Walk = batters$HR/batters$PA
percentages$SO = batters$SO/batters$AB


#Now let's look at our table:
percentages
#So a batter struck out about 24% of the time, got a single about 20% of the time, and homered about 4% of the time.

#Let's add the percentage of an out too (for now I am assuming no errors, HBP, SAC, etc..)
percentages$Out = 1 - percentages$Singles - percentages$Doubles - percentages$Triples - percentages$HR - percentages$Walk - percentages$SO

percentages #So a regular out occurs about 43% of the time. 

#Now that I have these percentages, I can randomly generate a number in between some of those percentages.
#Think of it like finding a random number inside of a confidence interval.
#For example, the average percentage of a single is about 20%, but suppose the range in the MLB is between
#15% and 25% (You can always adjust this). I can use the 'runif' command to generate a random number in the interval.

runif(1,.15,.25) #The 1 represents how many numbers I want R to spit out.
runif(3, 10,20)
#See if you get any interesting numbers. When I ran the interval (10,20) three times, I got three numbers
#all over 18. Interesting, right?

#Note that the distributions are uniform. Hence, I have the same chance of getting .223 as I am getting .159.

#So now let's create our batters. I will randomly generate 25 numbers in between these ranges for each stat.
#I will assign them to their own variable.
#Note: I would usually create actual Confidence Intervals for these numbers, but for a quick tutorial
#I will just use values that seem logical.
#Note that these values will be different for you than me.

Batter.Single = runif(25,.15,.25)
Batter.Single
Batter.Double = runif(25,.01,.09)
Batter.Double
Batter.Triple = runif(25,0,.01)
Batter.Triple
Batter.HR = runif(25, .005,.07)
Batter.HR
Batter.BB = runif(25, .005,.10)
Batter.BB
Batter.SO = runif(25,.10,.39)
Batter.SO

#Now let's put this all in a data frame.
BatterSplits = data.frame(Batter.Single,Batter.Double,Batter.Triple,Batter.HR,Batter.BB,Batter.SO)
colnames(BatterSplits) = c("Singles","Doubles","Triples","HR","Walks","SO")
BatterSplits

#Now let's add Outs at the end, the same way we did for percentages:
BatterSplits$Out = 1 - BatterSplits$Singles - BatterSplits$Doubles - BatterSplits$Triples - 
  BatterSplits$HR - BatterSplits$Walk - BatterSplits$SO


#OK, we finished with batters! Now we need to do the same for pitchers given what we want.
#I'll use the percentages of walks, SO, HR, Hits-HR, and Outs.

pitchers = pitchers[1,]
probSO = pitchers$SO/pitchers$BF # 0.2164333
probBB = pitchers$BB/pitchers$BF # 0.08542594
probHR = pitchers$HR/pitchers$BF # 0.03294746
probHit = pitchers$H/pitchers$BF - probHR # 0.1948784
probOut = 1- probSO - probBB - probHR - probHit # 0.4703149

randSO = runif(25,.10,.33)
randBB = runif(25,.02,.14)
randHR = runif(25,.01,.05)
randHit = runif(25,.14,.25) - randHR
randOut = 1 - randSO - randBB - randHR - randHit

PitcherSplits = data.frame(randSO, randBB, randHR, randHit, randOut)
colnames(PitcherSplits) = c("SO","BB","HR","H","Out")

#I am going to create another column, the "PitcherNumbers" column, which will determine
#if the numbers we want to use will belong the the hitter or the batter. I think this will be better
#explained once I create our function.
#Notice that I am not randomizing this, but basing it on how likely a pitcher is to get someone out.

PitcherSplits$PitcherNumber = ifelse(PitcherSplits$Out + PitcherSplits$SO > .75, .70,
       ifelse(PitcherSplits$Out+PitcherSplits$SO < .6, .50, .60))
#Basically, if the probability of getting an out or strikeout is > .75, they have a 75% chance
#Of directing all action to their numbers, not the batters numbers.


##So now I need to combine the two together and see what the result will be. I can create a function
#for this.
#A function will run anything you want when it is given the right arguments.
#I want this function to read in a Batters and Pitchers probabilities of getting a certain outcome.
#I'm gonna make a lot of if/else statements.
#A function will always need to RETURN something in an outcome. See what I do below and see if you
#can decipher it.



##These ifelse statements are saying "If the randomly generated number is in between this number
##and this number, then it is a "Single" or a "Double" etc. etc. 
batterVsPitcher <- function(Batter1B, Batter2B, Batter3B, BatterHR, BatterBB,
                            BatterSO, BatterOut, PitcherSO, PitcherBB, PitcherHR,
                            PitcherH, PitcherOut, PitcherNumbers) {
  
  #Generate a random number to decide whether it goes to the batter or the pitcher
  result = ""
  randNum = runif(1,0,1)
  if(randNum > PitcherNumbers) {
    randBatterNumber = runif(1,0,1) 
    result = ifelse(randBatterNumber < Batter1B, "Single",
                ifelse(randBatterNumber < Batter1B+Batter2B, "Double",
                     ifelse(randBatterNumber < Batter1B+Batter2B+Batter3B, "Triple",
                         ifelse(randBatterNumber < Batter1B+Batter2B+Batter3B+BatterHR, "Home Run",
                                ifelse(randBatterNumber < Batter1B+Batter2B+Batter3B+BatterHR+BatterBB, "Walk",
                                       ifelse(randBatterNumber < Batter1B+Batter2B+Batter3B+BatterHR+BatterBB+BatterSO, "Strikeout", "Out"))))))
              
    } else {
    randPitcherNumber = runif(1,0,1)
    #Notice that I adjusted the Pitcher's hits to only include singles. You can change this if you'd like.
    result = ifelse(randPitcherNumber < PitcherSO, "Strikeout",
                    ifelse(randPitcherNumber < PitcherSO+PitcherBB, "Walk",
                           ifelse(randPitcherNumber < PitcherSO+PitcherBB+PitcherHR,"Home Run",
                                  ifelse(randPitcherNumber < PitcherSO+PitcherBB+PitcherHR+PitcherH, "Single",
                                       "Out"))))
  }
  
  
  return(result)
}

##Now let's try our work.Remember that I have to add all of the arguments it asks for: Batter1B, Batter2B...PitcherNumbers

#Let's take the example of the first batter in our data vs the first pitcher.
batterVsPitcher(BatterSplits[1,1],BatterSplits[1,2],
                BatterSplits[1,3],BatterSplits[1,4],
                BatterSplits[1,5], BatterSplits[1,6],
                BatterSplits[1,7], PitcherSplits[1,1],
                PitcherSplits[1,2],PitcherSplits[1,3],
                PitcherSplits[1,4],PitcherSplits[1,5],
                PitcherSplits[1,6])

#Run this command a few times. See what you get!

#Now ask yourself the next step. How do you take these values and store them? How do you create
#a batting average? And finally, how to you adjust to find the best players?