Illustrate of bivariate density estimates on 2019 Statcast data

density_estimates.Rmd

---
title: "Some Density Estimates"
output:
  html_document:
    df_print: paged
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE,
                      message = FALSE,
                      warning = FALSE)
```

#### Ideas (2D density estimation via ggplot2)

- density estimation is very good for graphing large amounts of two-dimensional data

- better than scatterplots where one has to handle the overplotting issue

- maybe harder to interpret?

- basic function for computing density estimate is ```kde2d()```

- output of density estimate is a contour plot

- personally like filled contour bands which displays region of highest probability

- geometric object is ```geom_density2d_filled()```

- illustrate for (1) plotting locations in a zone, (2) in-play locations of batted balls, (3) distribution of launch variables

- when you have different facet's distinguish from joint density estimate or conditional density estimate for each facet (this ensures that peak intensity is the same in each facet)

- I think one is more interested in seeing conditional density estimates (one gets the conditional estimates by the contour_var = "ndensity" argument option)

For this exercise I will just be using the readr and ggplot2 and dplyr packages.

```{r}
library(readr)
library(ggplot2)
library(dplyr)
```

Load in Statcast data for the complete 2019 season:

```{r}
statcast2020 <- read_csv("~/Dropbox/2016 WORK/BLOG Baseball R/OTHER/StatcastData/statcast2019.csv")
```

Here are several functions that I use to cut down on typing the ggplot() functions.  I find the add_zone() especially helpful for adding the zone boundaries to the plot.

```{r}
increasefont <- function(){
  theme(text=element_text(size=18))
}
centertitle <- function(){
  theme(plot.title = element_text(colour = "blue", size = 18,
                hjust = 0.5, vjust = 0.8, angle = 0))
}
add_zone <- function(Color = "red"){
  topKzone <- 3.5
  botKzone <- 1.6
  inKzone <- -0.85
  outKzone <- 0.85
  kZone <- data.frame(
    x=c(inKzone, inKzone, outKzone, outKzone, inKzone),
    y=c(botKzone, topKzone, topKzone, botKzone, botKzone)
  )
  geom_path(aes(.data$x, .data$y), data=kZone,
              lwd=1, col=Color)
}
```


#### Locations of Pitches

Locations of different pitch types for right-handed pitchers:

```{r}
ggplot(filter(statcast2020,
              p_throws == "R",
              pitch_type %in%
            c("CH", "CS", "CU", "FC", "FF",
              "FS", "FT", "KC", "SI", "SL")),
       aes(plate_x, plate_z)) +
  geom_density2d_filled(contour_var = "ndensity") +
  add_zone("red") +
  xlim(-2.2, 2.2) +
  ylim(-0.5, 4) +
  facet_wrap(~ pitch_name) +
  ggtitle("Right-Arm Pitchers") +
  centertitle() +
  theme(text=element_text(size=16)) +
  theme(legend.position = "none") +
  coord_equal()
```

Locations of different pitch types for left-handed pitchers:

```{r}
ggplot(filter(statcast2020,
              p_throws == "L",
              pitch_type %in%
                c("CH", "CS", "CU", "FC", "FF",
                  "FS", "FT", "KC", "SI", "SL")),
       aes(plate_x, plate_z)) +
  geom_density2d_filled(contour_var = "ndensity") +
  add_zone("red") +
  xlim(-2.2, 2.2) +
  ylim(-0.5, 4) +
  facet_wrap(~ pitch_name) +
  ggtitle("Left-Arm Pitchers") +
  centertitle() +
  theme(text=element_text(size=16)) +
  theme(legend.position = "none") +
  coord_equal()
```

#### Zone locations of various hits

Zone locations of home runs

```{r}
ggplot(filter(statcast2020,
              events == "home_run"),
       aes(plate_x, plate_z)) +
  geom_density2d_filled() +
  add_zone("red") +
  xlim(-1.5, 1.5) +
  ylim(1, 4) +
  ggtitle("Home Runs") +
  centertitle() +
  facet_wrap(~ stand) +
  coord_equal()+
  theme(legend.position = "none")
```

Zone locations of hits

```{r}
ggplot(filter(statcast2020,
              events %in% c("single", "double",
                          "triple", "home_run")),
       aes(plate_x, plate_z)) +
  geom_density2d_filled() +
  add_zone("red") +
  xlim(-1.5, 1.5) +
  ylim(1, 4) +
  ggtitle("Hits") +
  centertitle() +
  facet_wrap(~ stand) +
  coord_equal()+
  theme(legend.position = "none")
```


Locations of hit locations by pitching arm and batter side

```{r}
ggplot(filter(statcast2020,
              events %in% c("single", "double",
                          "triple", "home_run")),
       aes(plate_x, plate_z)) +
  geom_density2d_filled(contour_var = "ndensity") +
  add_zone("red") +
  xlim(-1.5, 1.5) +
  ylim(1, 4) +
  ggtitle("Hits") +
  centertitle() +
  facet_grid(stand ~ p_throws,
             labeller=label_both) +
  increasefont() +
  theme(legend.position = "none") +
  coord_equal()
```

#### Locations of strikes

Locations of called strikes

```{r}
ggplot(filter(statcast2020,
              description == "called_strike"),
       aes(plate_x, plate_z)) +
  geom_density2d_filled() +
  add_zone("red") +
  xlim(-1.5, 1.5) +
  ylim(1, 4) +
  ggtitle("Called Strike") +
  centertitle() +
  facet_wrap(~ stand) +
  coord_equal()+
  theme(legend.position = "none")
```


Divide by pitching arm and batting side

```{r}
ggplot(filter(statcast2020,
              description == "called_strike"),
       aes(plate_x, plate_z)) +
  geom_density2d_filled(contour_var = "ndensity") +
  add_zone("red") +
  xlim(-1.5, 1.5) +
  ylim(1, 4) +
  ggtitle("Called Strike") +
  centertitle() +
  facet_grid(stand ~ p_throws,
             labeller=label_both) +
  increasefont() +
  theme(legend.position = "none") +
  coord_equal()
```

Location of swinging strikes

```{r}
ggplot(filter(statcast2020,
              description %in%
                c("swinging_strike",
                  "swinging_strike_blocked")),
       aes(plate_x, plate_z)) +
  geom_density2d_filled(contour_var = "ndensity") +
  add_zone("red") +
  xlim(-1.5, 1.5) +
  ylim(1, 4) +
  ggtitle("Swinging Strike") +
  centertitle() +
  facet_grid(stand ~ p_throws,
             labeller=label_both) +
  increasefont() +
  theme(legend.position = "none") +
  coord_equal()
```

#### Balls in Play

Field locations of BIP (note that I am transforming the variables hc_x and hc_y to get reasonable looking field locations):

```{r}
ggplot(filter(statcast2020,
              type == "X"),
       aes(hc_x - 125.42, 198.27 - hc_y)) +
  geom_density2d_filled() +
  coord_equal()+
  theme(legend.position = "none") +
  xlab("") + ylab("")
```

Locations of doubles

```{r}
ggplot(filter(statcast2020,
              type == "X",
              events == "double"),
       aes(hc_x - 125.42, 198.27 - hc_y)) +
  geom_density2d_filled() +
  coord_equal()+
  theme(legend.position = "none")+
  xlab("") + ylab("")
```

Locations of home runs

```{r}
ggplot(filter(statcast2020,
              type == "X",
              events == "home_run"),
       aes(hc_x - 125.42, 198.27 - hc_y)) +
  geom_density2d_filled() +
  coord_equal()+
  theme(legend.position = "none")+
  xlab("") + ylab("")
```

Look at locations of BIP types -- define a new variable BB_Type:

```{r}
statcast2020 %>%
  filter(type == "X") %>%
  mutate(BB_Type = ifelse(launch_angle <= 10,
                          "Ground Ball",
                   ifelse(launch_angle <= 25,
                          "Line Drive",
                   ifelse(launch_angle <= 50,
                          "Fly Ball",
                          "Pop Up")))) ->
  scip
```

Look at batters of each side -- locations of four types of batted balls.

```{r}
ggplot(filter(scip, stand == "L",
              is.na(BB_Type) == FALSE),
       aes(hc_x - 125.42, 198.27 - hc_y)) +
  geom_density2d_filled(contour_var = "ndensity") +
  ylim(-20, 80) +
  xlim(-50, 50) +
  ggtitle("Left-Handed Batters") +
  increasefont() +
  centertitle() +
  theme(legend.position = "none")+
  coord_equal() +
  xlab("") + ylab("") +
  facet_wrap(~ BB_Type, ncol = 2)
```

```{r}
ggplot(filter(scip, stand == "R",
              is.na(BB_Type) == FALSE),
       aes(hc_x - 125.42, 198.27 - hc_y)) +
  geom_density2d_filled(contour_var = "ndensity") +
  ylim(-20, 80) +
  xlim(-50, 50) +
  ggtitle("Right-Handed Batters") +
  increasefont() +
  centertitle() +
  theme(legend.position = "none")+
  coord_equal() +
  xlab("") + ylab("") +
  facet_wrap(~ BB_Type, ncol = 2) 
```

#### Distribution of launch variables launch angle and launch speed

All batted balls:

```{r}
ggplot(scip,
       aes(launch_angle, launch_speed)) +
  geom_density2d_filled() +
  ggtitle("Launch Variables") +
  increasefont() +
  centertitle() +
  theme(legend.position = "none") +
  xlim(-50, 100) +
  ylim(55, 120)
```

Home runs:

```{r}
ggplot(filter(scip, events == "home_run"),
       aes(launch_angle, launch_speed)) +
  geom_density2d_filled() +
  ggtitle("Launch Variables for Home Runs") +
  increasefont() +
  centertitle() +
  theme(legend.position = "none") +
  xlim(15, 42) +
  ylim(92, 112)
```