.Rhistory

"Scoring Average", "GIR") #Continuous variables used in analysis
#Loop checking the amount of NA in the dataset
for(var_DA in data_analysis){
na_check = sum(is.na(var_DA))
print(na_check)
}
#Mean imputation for the 2 missing values, so i can have a complete dataset
data_analysis$`Approaches > 100yds`[which(is.na(data_analysis$`Approaches > 100yds`))]=mean(data_analysis$`Approaches > 100yds`, na.rm = TRUE)
data_analysis$`Approaches < 100yds`[which(is.na(data_analysis$`Approaches < 100yds`))]=mean(data_analysis$`Approaches < 100yds`, na.rm = TRUE)
#Loop checking the amount of NA in the dataset post imputation
for(var_DA in data_analysis){
na_check = sum(is.na(var_DA))
print(na_check)
}
summary(data_analysis)
categories = names(data_analysis)
classes = sapply(data_analysis,class)
for(val in categories[classes == 'numeric'])
{
sd(data_analysis[,val])
}
sd
sd(data_analysis)
sd(data_analysis$`Driving Distance`)
sd(data_analysis$`Driving Distance`)
sd(data_analysis$`Driving Accuracy %`)
sd(data_analysis$`Approaches < 100yds`)
sd(data_analysis$`Approaches > 100yds`)
sd(data_analysis$GIR)
sd(data_analysis$`Bogey Average`)
sd(data_analysis$`Scoring Average`)
summary(data_analysis)
sd(data_analysis$`Driving Distance`)
sd(data_analysis$`Driving Accuracy %`)
sd(data_analysis$`Approaches < 100yds`)
sd(data_analysis$`Approaches > 100yds`)
sd(data_analysis$GIR)
sd(data_analysis$`Bogey Average`)
sd(data_analysis$`Scoring Average`)
plot(data_analysis$`Driving Distance`)
plot(data_analysis$`Driving Distance`,data_analysis$`Driving Accuracy %`)
pairs(~`Driving Distance`+`Driving Accuracy %`+`Approaches < 100yds`+
`Approaches < 100yds`+`Approaches > 100yds`+GIR+`Bogey Average`+
`Scoring Average`)
pairs(~`Driving Distance`+`Driving Accuracy %`+`Approaches < 100yds`+
`Approaches < 100yds`+`Approaches > 100yds`+GIR+`Bogey Average`+
`Scoring Average`, data = data_analysis)
packages = c("mice","readr","rvest","purrr","dplyr", "Hmisc","ggplot2",
"reshape2","lsr","caTools","heplots","jtools","ggstance",
"broom.mixed","fmsb","scales","tidyverse")
lapply(packages, require, character.only = TRUE)#Loading all required packages
packages = c("mice","readr","rvest","purrr","dplyr", "Hmisc","ggplot2",
"reshape2","lsr","caTools","heplots","jtools","ggstance",
"broom.mixed","fmsb","scales","tidyverse")
lapply(packages, require, character.only = TRUE)#Loading all required packages
#Data retrieval
data = read.csv(file="pgadata.csv") #Importing data from CSV
ls(data) #To see the variables in the dataset
#Keeping only the variables of interest
data_analysis = subset(data, select = c(Player,
Year,
Driving.Distance,
Driving.Accuracy.Percentage,
Approaches.from.less.than.100.yards,
Approaches.from.greater.than.100.yards,
GIR,
Top.10.Finishes,
Bogey.Average,
Scoring.Average))
str(data_analysis) #To verify the variable type
#Renaming variable names so they don't have periods
col_name = c("Driving.Distance", "Driving.Accuracy.Percentage",
"Approaches.from.less.than.100.yards",
"Approaches.from.greater.than.100.yards",
"Top.10.Finishes","Bogey.Average", "Scoring.Average")
new_col_name = c("Driving Distance", "Driving Accuracy %","Approaches < 100yds",
"Approaches > 100yds","Top 10 Finishes","Bogey Average",
"Scoring Average")
#Renaming column names to shorten lines of code
colnames(data_analysis)[colnames(data_analysis) %in% col_name] = new_col_name
#I was missing the APLH variable from CSV dataset, thus I webscraped it from the PGA tour website
link = "https://www.pgatour.com/stats/stat.02329.y%d.html"
APLH_data = map_df(2015:2019, function(i){
page = read_html(sprintf(link,i))
data.frame(Player = html_text(html_nodes(page, "#statsTable a")),
APLH = html_text(html_nodes(page, "td:nth-child(5)"))
)
})
#Each year has a different amount of data point, this is to account for it
APLH_data['Year'] = rep(2015:2019, c(184,185,190,193,188))
#Data Cleaning
#Created a function to delete n missing values
delete.na = function(data_analysis, n=0) {
data_analysis[rowSums(is.na(data_analysis)) <= n,]
}
#Deleting players who are missing at least 6 data points out of the 7 continuous variables.
#To many missing data point to impute.
data_analysis = delete.na(data_analysis, 6)
#Replacing NA values in Top 10 finishes with 0.
data_analysis[c("Top 10 Finishes")][is.na(data_analysis[c("Top 10 Finishes")])] = 0
#Checking is there are any NA in the dataset of intrest
var_DA = c("Driving Distance", "Driving Accuracy %","Approaches < 100yds",
"Approaches > 100yds","Top 10 Finishes","Bogey Average",
"Scoring Average", "GIR") #Continuous variables used in analysis
#Loop checking the amount of NA in the dataset
for(var_DA in data_analysis){
na_check = sum(is.na(var_DA))
print(na_check)
}
#Mean imputation for the 2 missing values, so i can have a complete dataset
data_analysis$`Approaches > 100yds`[which(is.na(data_analysis$`Approaches > 100yds`))]=mean(data_analysis$`Approaches > 100yds`, na.rm = TRUE)
data_analysis$`Approaches < 100yds`[which(is.na(data_analysis$`Approaches < 100yds`))]=mean(data_analysis$`Approaches < 100yds`, na.rm = TRUE)
#Loop checking the amount of NA in the dataset post imputation
for(var_DA in data_analysis){
na_check = sum(is.na(var_DA))
print(na_check)
}
#Descriptive statistics of Data
summary(data_analysis)
sd(data_analysis$`Driving Distance`)
sd(data_analysis$`Driving Accuracy %`)
sd(data_analysis$`Approaches < 100yds`)
sd(data_analysis$`Approaches > 100yds`)
sd(data_analysis$GIR)
sd(data_analysis$`Bogey Average`)
sd(data_analysis$`Scoring Average`)
split = sample.split(data_analysis$GIR, SplitRatio = 0.7)
#Setting up traing and test dataset
training_set = subset(data_analysis, split == TRUE)
test_set = subset(data_analysis, split == FALSE)
#Please see reference list under Ford. C for full reference (beginning)
multiregmodel_training = lm(cbind(GIR, `Bogey Average`, `Scoring Average`) ~
`Driving Distance` +
`Driving Accuracy %` +
`Approaches < 100yds` +
`Approaches > 100yds`,
data = training_set)
#End of Ford. C reference
summary(multiregmodel_training)#To determine the estimates
summary.aov(multiregmodel_training) #To determine the SSR,SSE and SSTO
etasq(multiregmodel_training)#Variation accounted for by each IV for the entire model
pred_values_MMLR = predict(multiregmodel_training, newdata = test_set)#Predicting the DV using the test data set
test_set$pred_values_MMLR = pred_values_MMLR#Including the predicted DV into the test data set
#Renaming the Predicted DV values
test_set$Predicted_GIR = pred_values_MMLR[,"GIR"]
test_set$Predicted_BA = pred_values_MMLR[,"Bogey Average"]
test_set$Predicted_SA = pred_values_MMLR[,"Scoring Average"]
col_name_test_set = c("Predicted_GIR", "Predicted_BA", "Predicted_SA")
new_col_name_test_set = c("Predicted GIR", "Predicted Bogey Average", "Predicted Scoring Average")
colnames(test_set)[colnames(test_set) %in% col_name_test_set] = new_col_name_test_set
#Plotting the observed points against predicted points, the regression line
plot(test_set$GIR, test_set$`Predicted GIR`)
abline(lm(`Predicted GIR`~GIR, data = test_set))
plot(test_set$`Bogey Average`, test_set$`Predicted Bogey Average`)
abline(lm(`Predicted Bogey Average`~`Bogey Average`, data = test_set))
plot(test_set$`Scoring Average`, test_set$`Predicted Scoring Average`)
abline(lm(`Predicted Scoring Average`~`Scoring Average`, data = test_set))
#Mean absolute percentage error formula
MAPE_formula_GIR =
abs(mean(((test_set$GIR - test_set$`Predicted GIR`)/(test_set$GIR))))*100
MAPE_formula_BA =
abs(mean(((test_set$`Bogey Average` - test_set$`Predicted Bogey Average`)/
(test_set$`Bogey Average`))))*100
MAPE_formula_SA =
abs(mean(((test_set$`Scoring Average` - test_set$`Predicted Scoring Average`)/
(test_set$`Scoring Average`))))*100
#Different visualizations for regression output and estimates
m1 = lm(GIR ~`Driving Distance` + `Driving Accuracy %` + `Approaches < 100yds` +
`Approaches > 100yds`, data = training_set)
m2 = lm(`Bogey Average` ~ `Driving Distance` + `Driving Accuracy %` +
`Approaches < 100yds` +
`Approaches > 100yds`, data = training_set)
m3 = lm(`Scoring Average` ~`Driving Distance` + `Driving Accuracy %` +
`Approaches < 100yds` +
`Approaches > 100yds`, data = training_set)
summ(m1, confint = TRUE, digits = 3)
summ(m2, confint = TRUE, digits = 3)
summ(m3, confint = TRUE, digits = 3)
#Setting up a dataframe to hold the values of model accuracy, MAPE and adjR^2
Model = c("Model 1: GIR", "Model 2: Bogey Average", "Model 3: Scoring Average")
MAPE.Percentage = c(MAPE_formula_GIR,MAPE_formula_BA,MAPE_formula_SA)
adjR.sqrd = c(summary(m1)$adj.r.squared,summary(m2)$adj.r.squared,
summary(m3)$adj.r.squared)
table = data.frame(Model,MAPE.Percentage,adjR.sqrd)
print(table)
#Variation accounted for by each IV for the each respective DV
etaSquared(m1)
etaSquared(m2)
etaSquared(m3)
#Normality and leverage plots for each regression
plot(m1)
#Plotting Regression estimators
#Please see reference list under Long. J for full reference (beginning)
plot_summs(m1,m2,m3,inner_ci_level = 0.95,
model.names = c("GIR", "Bogey Average", "Scoring Average"),
legend.title = "Models", point.size = 5)
#End of Long. J reference
#Installing and loading required packages
install.packages(c("mice","readr","rvest","purrr","dplyr", "Hmisc","ggplot2",
"reshape2","lsr","caTools","heplots","jtools","ggstance",
"broom.mixed","fmsb","scales","tidyverse")) #Installing required packages
#Nikolas Argiropoulos, 40044358
#STAT 497 - Sports Analytics Final Project Code
#Installing and loading required packages
install.packages(c("mice","readr","rvest","purrr","dplyr", "Hmisc","ggplot2",
"reshape2","lsr","caTools","heplots","jtools","ggstance",
"broom.mixed","fmsb","scales","tidyverse")) #Installing required packages
packages = c("mice","readr","rvest","purrr","dplyr", "Hmisc","ggplot2",
"reshape2","lsr","caTools","heplots","jtools","ggstance",
"broom.mixed","fmsb","scales","tidyverse")
lapply(packages, require, character.only = TRUE)#Loading all required packages
#Data retrieval
data = read.csv(file="pgadata.csv") #Importing data from CSV
ls(data) #To see the variables in the dataset
#Keeping only the variables of interest
data_analysis = subset(data, select = c(Player,
Year,
Driving.Distance,
Driving.Accuracy.Percentage,
Approaches.from.less.than.100.yards,
Approaches.from.greater.than.100.yards,
GIR,
Top.10.Finishes,
Bogey.Average,
Scoring.Average))
str(data_analysis) #To verify the variable type
#Renaming variable names so they don't have periods
col_name = c("Driving.Distance", "Driving.Accuracy.Percentage",
"Approaches.from.less.than.100.yards",
"Approaches.from.greater.than.100.yards",
"Top.10.Finishes","Bogey.Average", "Scoring.Average")
new_col_name = c("Driving Distance", "Driving Accuracy %","Approaches < 100yds",
"Approaches > 100yds","Top 10 Finishes","Bogey Average",
"Scoring Average")
#Renaming column names to shorten lines of code
colnames(data_analysis)[colnames(data_analysis) %in% col_name] = new_col_name
#I was missing the APLH variable from CSV dataset, thus I webscraped it from the PGA tour website
link = "https://www.pgatour.com/stats/stat.02329.y%d.html"
APLH_data = map_df(2015:2019, function(i){
page = read_html(sprintf(link,i))
data.frame(Player = html_text(html_nodes(page, "#statsTable a")),
APLH = html_text(html_nodes(page, "td:nth-child(5)"))
)
})
#Each year has a different amount of data point, this is to account for it
APLH_data['Year'] = rep(2015:2019, c(184,185,190,193,188))
#Data Cleaning
#Created a function to delete n missing values
delete.na = function(data_analysis, n=0) {
data_analysis[rowSums(is.na(data_analysis)) <= n,]
}
#Deleting players who are missing at least 6 data points out of the 7 continuous variables.
#To many missing data point to impute.
data_analysis = delete.na(data_analysis, 6)
#Replacing NA values in Top 10 finishes with 0.
data_analysis[c("Top 10 Finishes")][is.na(data_analysis[c("Top 10 Finishes")])] = 0
#Checking is there are any NA in the dataset of intrest
var_DA = c("Driving Distance", "Driving Accuracy %","Approaches < 100yds",
"Approaches > 100yds","Top 10 Finishes","Bogey Average",
"Scoring Average", "GIR") #Continuous variables used in analysis
#Loop checking the amount of NA in the dataset
for(var_DA in data_analysis){
na_check = sum(is.na(var_DA))
print(na_check)
}
#Mean imputation for the 2 missing values, so i can have a complete dataset
data_analysis$`Approaches > 100yds`[which(is.na(data_analysis$`Approaches > 100yds`))]=mean(data_analysis$`Approaches > 100yds`, na.rm = TRUE)
data_analysis$`Approaches < 100yds`[which(is.na(data_analysis$`Approaches < 100yds`))]=mean(data_analysis$`Approaches < 100yds`, na.rm = TRUE)
#Loop checking the amount of NA in the dataset post imputation
for(var_DA in data_analysis){
na_check = sum(is.na(var_DA))
print(na_check)
}
#Descriptive statistics of Data
summary(data_analysis)
sd(data_analysis$`Driving Distance`)
sd(data_analysis$`Driving Accuracy %`)
sd(data_analysis$`Approaches < 100yds`)
sd(data_analysis$`Approaches > 100yds`)
sd(data_analysis$GIR)
sd(data_analysis$`Bogey Average`)
sd(data_analysis$`Scoring Average`)
#Normality Tests
#Scatter plots for linear assumption of regression
pairs(~`Driving Distance`+`Driving Accuracy %`+`Approaches < 100yds`+
`Approaches < 100yds`+`Approaches > 100yds`+GIR+`Bogey Average`+
`Scoring Average`, data = data_analysis)
#Creating histogram for dataset, density not frequency
categories = names(data_analysis)
classes = sapply(data_analysis,class)
for(val in categories[classes == 'numeric'])
{
hist(data_analysis[,val], main= paste("Historgram of", val),col='steelblue',
xlab = paste(val), prob = TRUE) # subset with [] not $
lines(density(data_analysis[, val]), col = "red", lwd = 3)
}
#Creating Q-Q plot for dataset
for(val in categories[classes == 'numeric'])
{
qqnorm(data_analysis[,val], main= paste("QQPlot of", val),col='steelblue', xlab = paste(val))
qqline(data_analysis[,val], col = "red", lwd = 2)
}
#Performing shapiro-wilk test
for (val in  categories[classes == 'numeric']){
NT = shapiro.test(data_analysis[,val])
print(NT)
}
#Only continuous data of interest
cont_data = subset(data_analysis, select = c(`Driving Distance`,
`Driving Accuracy %`,
`Approaches < 100yds`,
`Approaches > 100yds`,
GIR,`Top 10 Finishes`,
`Bogey Average`,`Scoring Average`))
#Correlation matrix to set up ggplot
cor_matrix = round(cor(cont_data),2)
#Correlations in numerical format
corrdata = melt(cor_matrix)
#Correlation matrix
ggplot(corrdata, aes(x = Var1, y = Var2, fill = value)) +
geom_tile() +
scale_fill_gradient2(midpoint = 0.5, low = "cyan",mid = "deepskyblue4",
limits = c(-1, +1)) +
labs(title = "Correlation Matrix \n of Golf KPI \n",
x = "", y = "", fill = "Correlation \n Coefficient") +
theme(plot.title = element_text(hjust = 0.5, colour = "blue", size = 12,face = "bold"),
legend.title = element_text(face="bold", colour="blue", size = 12),
axis.text.x = element_text(angle = 45, hjust=1, size = 12),
axis.text.y = element_text(size = 12)) +
geom_text(aes(x = Var1, y = Var2, label = round(value, 2)),
color = "black", fontface = "bold", size = 5)
#Multiple regression analysis
#Splitting the GIR DV, no need to split the other 2 since the entire dataset gets splits
split = sample.split(data_analysis$GIR, SplitRatio = 0.7)
#Setting up traing and test dataset
training_set = subset(data_analysis, split == TRUE)
test_set = subset(data_analysis, split == FALSE)
#Please see reference list under Ford. C for full reference (beginning)
multiregmodel_training = lm(cbind(GIR, `Bogey Average`, `Scoring Average`) ~
`Driving Distance` +
`Driving Accuracy %` +
`Approaches < 100yds` +
`Approaches > 100yds`,
data = training_set)
#End of Ford. C reference
summary(multiregmodel_training)#To determine the estimates
summary.aov(multiregmodel_training) #To determine the SSR,SSE and SSTO
etasq(multiregmodel_training)#Variation accounted for by each IV for the entire model
pred_values_MMLR = predict(multiregmodel_training, newdata = test_set)#Predicting the DV using the test data set
test_set$pred_values_MMLR = pred_values_MMLR#Including the predicted DV into the test data set
#Renaming the Predicted DV values
test_set$Predicted_GIR = pred_values_MMLR[,"GIR"]
test_set$Predicted_BA = pred_values_MMLR[,"Bogey Average"]
test_set$Predicted_SA = pred_values_MMLR[,"Scoring Average"]
col_name_test_set = c("Predicted_GIR", "Predicted_BA", "Predicted_SA")
new_col_name_test_set = c("Predicted GIR", "Predicted Bogey Average", "Predicted Scoring Average")
colnames(test_set)[colnames(test_set) %in% col_name_test_set] = new_col_name_test_set
#Plotting the observed points against predicted points, the regression line
plot(test_set$GIR, test_set$`Predicted GIR`)
abline(lm(`Predicted GIR`~GIR, data = test_set))
plot(test_set$`Bogey Average`, test_set$`Predicted Bogey Average`)
abline(lm(`Predicted Bogey Average`~`Bogey Average`, data = test_set))
plot(test_set$`Scoring Average`, test_set$`Predicted Scoring Average`)
abline(lm(`Predicted Scoring Average`~`Scoring Average`, data = test_set))
#Mean absolute percentage error formula
MAPE_formula_GIR =
abs(mean(((test_set$GIR - test_set$`Predicted GIR`)/(test_set$GIR))))*100
MAPE_formula_BA =
abs(mean(((test_set$`Bogey Average` - test_set$`Predicted Bogey Average`)/
(test_set$`Bogey Average`))))*100
MAPE_formula_SA =
abs(mean(((test_set$`Scoring Average` - test_set$`Predicted Scoring Average`)/
(test_set$`Scoring Average`))))*100
#Different visualizations for regression output and estimates
m1 = lm(GIR ~`Driving Distance` + `Driving Accuracy %` + `Approaches < 100yds` +
`Approaches > 100yds`, data = training_set)
m2 = lm(`Bogey Average` ~ `Driving Distance` + `Driving Accuracy %` +
`Approaches < 100yds` +
`Approaches > 100yds`, data = training_set)
m3 = lm(`Scoring Average` ~`Driving Distance` + `Driving Accuracy %` +
`Approaches < 100yds` +
`Approaches > 100yds`, data = training_set)
summ(m1, confint = TRUE, digits = 3)
summ(m2, confint = TRUE, digits = 3)
summ(m3, confint = TRUE, digits = 3)
#Setting up a dataframe to hold the values of model accuracy, MAPE and adjR^2
Model = c("Model 1: GIR", "Model 2: Bogey Average", "Model 3: Scoring Average")
MAPE.Percentage = c(MAPE_formula_GIR,MAPE_formula_BA,MAPE_formula_SA)
adjR.sqrd = c(summary(m1)$adj.r.squared,summary(m2)$adj.r.squared,
summary(m3)$adj.r.squared)
table = data.frame(Model,MAPE.Percentage,adjR.sqrd)
print(table)
#Variation accounted for by each IV for the each respective DV
etaSquared(m1)
etaSquared(m2)
etaSquared(m3)
#Normality and leverage plots for each regression
plot(m1)
#Plotting Regression estimators
#Please see reference list under Long. J for full reference (beginning)
plot_summs(m1,m2,m3,inner_ci_level = 0.95,
model.names = c("GIR", "Bogey Average", "Scoring Average"),
legend.title = "Models", point.size = 5)
#End of Long. J reference
#Radar chart for min, max and means
#Please see reference list under Alboukadel for full reference (beginning)
#Radar function
radarchart_combine = function(data, color = "#00AFBB",
vlabels = colnames(data), vlcex = 0.7,
caxislabels = NULL, title = NULL, ...){
radarchart(
data, axistype = 1,
# Customize the polygon
pcol = color, pfcol = scales::alpha(color, 0.5), plwd = 2, plty = 1,
# Customize the grid
cglcol = "grey", cglty = 1, cglwd = 0.8,
# Customize the axis
axislabcol = "grey",
# Variable labels
vlcex = vlcex, vlabels = vlabels,
caxislabels = caxislabels, title = title, ...
)
}
#Setting up data for the radar graph under radar chart function
radar_chart_data = data.frame(
row.names = c("Tour Average", "Rory McIlroy","Dustin Johnson", "Jordan Spieth"),
GIR = c(76.85,79.81, 80.19, 79.41),
Bogey.Average = c(2.61,2.29, 2.26, 2.22),
Scoring.Average = c(70.95 ,69.38, 69.38, 69.53),
Driving.Distance = c(293.10,314.30, 314.46, 292.56),
Driving.Accuracy.Percentage = c(61.37,58.43, 56.94, 59.14),
Approaches.from.less.than.100.yards = c(17.14,15.90, 15.48, 16.72),
Approaches.from.greater.than.100.yards = c(32.65,30.78, 30.52, 31.86),
Top.10.Finishes = c(2.73,8.75, 10.60, 8.80))
max_min = data.frame(
row.names = c("Max","Min"),
GIR = c(84.86, 65.15),
Bogey.Average = c(3.940, 1.860),
Scoring.Average = c(74.40, 68.70),
Driving.Distance = c(319.70, 269.70),
Driving.Accuracy.Percentage = c(76.88, 43.02),
Approaches.from.less.than.100.yards = c(27.30, 11.60),
Approaches.from.greater.than.100.yards = c(40.70, 27.50),
Top.10.Finishes = c(15.00, 0.00))
df = rbind(max_min,radar_chart_data)
#End of Alboukadel reference
#Renaming column names to shorten lines of code
colnames(df)[colnames(df) %in% col_name] = new_col_name
#Individual radar graphs
colors = c("darkorchid1", "dodgerblue1", "firebrick", "deeppink")
players = c("Tour Average","Rory McIlroy","Dustin Johnson", "Jordan Spieth")
for(i in 1:4){
radarchart_combine(
data = df[c(1, 2, i+2), ],
color = colors[i], title = players[i]
)
}
#Data preparation for radar chart comparing top three players against the tour average
Players = factor(c("Tour Average", "Rory McIlroy", "Dustin Johnson", "Jordan Spieth"),
levels = c("Tour Average", "Rory McIlroy", "Dustin Johnson", "Jordan Spieth"))
var_int = data.frame(GIR = c(76.85,79.81, 80.19, 79.41),
Bogey.Average = c(2.61,2.29, 2.26, 2.22),
Scoring.Average = c(70.95 ,69.38, 69.38, 69.53),
Driving.Distance = c(293.10,314.30, 314.46, 292.56),
Driving.Accuracy.Percentage = c(61.37,58.43, 56.94, 59.14),
Approaches.from.less.than.100.yards = c(17.14,15.90, 15.48, 16.72),
Approaches.from.greater.than.100.yards = c(32.65,30.78, 30.52, 31.86),
Top.10.Finishes = c(2.73,8.75, 10.60, 8.80))
#Re-scaling the data to be between 0 and 1
df_scaled = round(apply(var_int, 2, scales::rescale),2)
df_scaled = as.data.frame(df_scaled)
#Setting up the new dataframe
df_radar_plot = data.frame(Players = Players,
GIR = df_scaled$GIR,
Bogey.Average = df_scaled$Bogey.Average,
Scoring.Average = df_scaled$Scoring.Average,
Top_10_Finishes = df_scaled$Top.10.Finishes,
Driving_Distance = df_scaled$Driving.Distance,
Drive_Accuracy = df_scaled$Driving.Accuracy.Percentage,
Approaches_less_100yards = df_scaled$Approaches.from.less.than.100.yards,
Approaches_greater_100yards = df_scaled$Approaches.from.greater.than.100.yards)
#Combining and categorising the data
df_radar_plot = melt(df_radar_plot,
id.vars = c("Players"),
measure.vars = c("GIR", "Bogey.Average","Scoring.Average",
"Top_10_Finishes","Driving_Distance",
"Drive_Accuracy","Approaches_less_100yards",
"Approaches_greater_100yards"),
variable.name = "KPI",
value.name = "Percentage")
levels(df_radar_plot$KPI) = list(
GIR = "GIR",
`Bogey Average` = "Bogey.Average",
`Scoring Average` = "Scoring.Average",
`Top 10 Finishes` = "Top_10_Finishes",
`Driving Distance` = "Driving_Distance",
`Driving Accuracy %` = "Drive_Accuracy",
`Approaches < 100yds` = "Approaches_less_100yards",
`Approaches > 100yds` = "Approaches_greater_100yards")
#Radar chart for comparing top three players against the tour average
ggplot(data = df_radar_plot,
aes(x = KPI, y = Percentage, group = Players, colour = Players)) +
geom_polygon(linewidth = 1, alpha= 0.1, fill = "white") +
ggtitle("Top 3 PGA Tour Players compared to Tour Average")  +
theme(plot.title = element_text(hjust = 0.5)) +
ylim(-1.0, 1.0) +
scale_x_discrete() +
theme_light() +
scale_color_manual(values = c("Black", "Blue","Red","Green")) +
scale_fill_manual(values = c("Black", "Blue","Red","Green")) +
coord_polar()
install.packages(c("mice", "readr", "rvest", "purrr", "dplyr", "Hmisc", "ggplot2", "reshape2", "lsr", "caTools", "heplots", "jtools", "ggstance", "broom.mixed", "fmsb", "scales", "tidyverse"))