library(ggplot2) # plot your data
library(tidyr) # tidy your data
library(lubridate) # time conversion
library(plyr) # rename single columns of data frames

# ggplot layout
legend_theme <- theme(plot.title = element_text(size=20, hjust = 0.5, face="bold"), 
                      axis.title.x = element_text(size=14, face="bold"),
                      axis.title.y = element_text(size=14, face="bold"),
                      legend.title=element_text(size=14),
                      legend.text=element_text(size=12))

# read data
data <- read.csv("soccer.csv", header=FALSE)

# fix column names
colnames(data) <- c("timestamp","tag_id","x_pos","y_pos","heading","direction","energy","speed","total_distance")

# convert date to date and seconds since game start
data <- data %>% separate(timestamp, into = c("date", "time"), sep=" ")
data$time <- as.numeric(hms(data$time))
data$time <- data$time-data$time[1]

# get player 2 data
player2 <- subset(data, tag_id == 2, select = c("time", "x_pos", "y_pos", "speed", "total_distance"))
# compute total distance
player2$dist2 <- c(0, cumsum(sqrt(diff(player2$x_pos)*diff(player2$x_pos)+diff(player2$y_pos)*diff(player2$y_pos))))
# adjust total distance from data file (substract offset)
player2$total_distance2 <- player2$total_distance - player2$total_distance[1]
# compute speed
player2$cspeed2 <- c(0, diff(player2$dist2)/diff(player2$time))
# rename speed from data file
player2 <- rename(player2, c("speed"="speed2"))

# same as above for player 5
player5 <- subset(data, tag_id == 5, select = c("time", "x_pos", "y_pos", "speed", "total_distance"))
player5$dist5 <- c(0, cumsum(sqrt(diff(player5$x_pos)*diff(player5$x_pos)+diff(player5$y_pos)*diff(player5$y_pos))))
player5$total_distance5 <- player5$total_distance - player5$total_distance[1]
player5$cspeed5 <- c(0, diff(player5$dist5)/diff(player5$time))
player5 <- rename(player5, c("speed"="speed5"))

# same as above for player 10
player10 <- subset(data, tag_id == 10, select = c("time", "x_pos", "y_pos", "speed", "total_distance"))
player10$dist10 <- c(0, cumsum(sqrt(diff(player10$x_pos)*diff(player10$x_pos)+diff(player10$y_pos)*diff(player10$y_pos))))
player10$total_distance10 <- player10$total_distance - player10$total_distance[1]
player10$cspeed10 <- c(0, diff(player10$dist10)/diff(player10$time))
player10 <- rename(player10, c("speed"="speed10"))

# "gather" value colums for distance/speed
player2 <- player2 %>% gather(type, dist, c("dist2","total_distance2")) %>% gather(stype, sspeed, c("cspeed2","speed2"))
player5 <- player5 %>% gather(type, dist, c("dist5","total_distance5")) %>% gather(stype, sspeed, c("cspeed5","speed5"))
player10 <- player10 %>% gather(type, dist, c("dist10","total_distance10")) %>% gather(stype, sspeed, c("cspeed10","speed10"))

# combine values into one dataframe
player2510 <- rbind(player2, player5, player10)

# convert time from seconds to minutes and distance from meter to kilometer
player2510$time = player2510$time/60
player2510$dist = player2510$dist/1000

# # convert time from seconds to minutes
# player2$time = player2$time/60
# player5$time = player5$time/60
# player10$time = player10$time/60
# 
# # convert distance from meter to kilometer
# player2$dist = player2$dist/1000
# player5$dist = player5$dist/1000
# player10$dist = player10$dist/1000

# plot total distance
gg <- ggplot()
gg <- gg + geom_line(data = player2510, aes(x=time, y=dist, color=type)) # single data frame
# gg <- gg + geom_line(data = player2, aes(x=time, y=dist, color=type)) # three data frames (1)
# gg <- gg + geom_line(data = player5, aes(x=time, y=dist, color=type)) # three data frames (2)
# gg <- gg + geom_line(data = player10, aes(x=time, y=dist, color=type)) # three data frames (3)
gg <- gg + labs(title = "Distance", x = "Time (min)", y = "Distance (km)") # rename labels
gg <- gg + scale_color_manual("Distance", # rename legend
                              values = c("dist2" = "yellow", "total_distance2" = "red", "dist5" = "blue", "total_distance5" = "black","dist10" = "green", "total_distance10" = "cyan"), # select colors
                              labels = c("dist2" = "Dist 2", "total_distance2" = "Total 2", "dist5" = "Dist 5", "total_distance5" = "Total 5","dist10" = "Dist 10", "total_distance10" = "Total 10")) # rename legend entries
gg <- gg + legend_theme;
print(gg)

# plot speed (similar to distance)
gg <- ggplot()
gg <- gg + geom_line(data = player2510, aes(x=time, y=sspeed, color=stype))
# gg <- gg + geom_line(data = player2, aes(x=time, y=sspeed, color=stype))
# gg <- gg + geom_line(data = player5, aes(x=time, y=sspeed, color=stype))
# gg <- gg + geom_line(data = player10, aes(x=time, y=sspeed, color=stype))
gg <- gg + labs(title = "Speed", x = "Time (min)", y = "Speed (m/s)")
gg <- gg + scale_color_manual("Speed", 
                              values = c("cspeed2" = "yellow", "speed2" = "red", "cspeed5" = "blue", "speed5" = "black","cspeed10" = "green", "speed10" = "cyan"))
gg <- gg + legend_theme;
print(gg)