library(tidyverse)
library(lubridate)
library(dplyr)
library(ggplot2)
library(leaflet)
library(shiny)
library(shinydashboard)

df <- read.csv(file = "crime.csv", header = TRUE, sep=",")

#remove missing values
df <- na.omit(df)

#Preprocessing
df$YEAR <- as.factor(df$YEAR)
df$HOUR <- as.factor(df$HOUR)
df$MONTH <- as.factor(df$MONTH)
df$OCCURRED_ON_DATE <- as.POSIXct(df$OCCURRED_ON_DATE, format = "%Y-%m-%d %H:%M:%OS")





row1 <- fluidRow( 
  
  tabBox(
    
    title = "Crime Frequency",
    id = "tabset1", height = "250px",
    
    tabPanel("Per Month", plotOutput("permonth")),
    tabPanel("Per Weekday", plotOutput("perday")),
    tabPanel("Per Hour", plotOutput("perhour"))
  )
  
  ,box(title = "Map View", solidHeader = TRUE 
       ,collapsible = TRUE, leafletOutput("mymap")) 
  
)

#Creating another fluid row
row2 <- fluidRow( 
  box(title = "Stacked Plot", solidHeader = TRUE 
      ,collapsible = TRUE, plotOutput("stacked"))
  
  ,box(title = "Timeseries Plot", solidHeader = TRUE 
       ,collapsible = TRUE, plotOutput("timeseries")) 
)


## Project Boston Crimes: UI - Header & Sidebars
header <- dashboardHeader(title = "Boston Crime Dataset")

sidebar <- dashboardSidebar(
  radioButtons("year", "Select Year: ",
               c("2015" = "2015",
                 "2016" = "2016",
                 "2017" = "2017",
                 "2018" = "2018",
                 "2019" = "2019")),
  
  selectInput("offense",
              "Offense Group:",
              c(unique(as.character(df$OFFENSE_CODE_GROUP[1:10])))),
  
  
  #Addition of a side bar menu to navigate to different tabs
  sidebarMenu(
    menuItem("Plots", tabName = "plots"),
    menuItem("Table", tabName = "table")
  )
  
)

#Adding components to body:


body <- dashboardBody(
  tabItems(
    tabItem("plots", row1, row2),
    tabItem("table", dataTableOutput('table'))
))


#Adding all components to UI:
ui <- dashboardPage(header, sidebar, body)



## Project Boston Crimes: Server
server <- function(input, output) { 
  
  #creating a reactive object
  filtered <- reactive({
    filter(filter(df, YEAR == input$year), OFFENSE_CODE_GROUP == input$offense)
  })
  
  #creating the plotOutput content
  output$permonth <- renderPlot({
    ggplot(data = filtered(), aes(x = filtered()$MONTH)) + geom_bar(fill="steelblue") + xlab("Month") + ylab("Number of Crimes") + theme_minimal()
  })
  
  output$perday <- renderPlot({
    ggplot(data = filtered(), aes(x = filtered()$DAY_OF_WEEK)) + geom_bar(fill="red") + xlab("Day of Week") + ylab("Number of Crimes") + theme_minimal()
  })
  
  output$perhour <- renderPlot({
    ggplot(data = filtered(),aes(x = filtered()$HOUR)) + geom_bar(fill="green") + xlab("Hour") + ylab("Number of Crimes") + theme_minimal()
  })
  
  #Defining the timeseries plot
  output$timeseries <- renderPlot({
    
    ggplot(data = filtered(), aes(x = filtered()$OCCURRED_ON_DATE))+ xlab("Date") + ylab("Number of Crimes") + geom_freqpoly(colour="purple")
  })
  
  #Defining the newly created stacked plot
  output$stacked <- renderPlot({
    
    filtered() %>% filter(DISTRICT %in% (filtered()  %>% count(DISTRICT) %>% arrange(-n) %>% head(7) %>% pull(DISTRICT)),
                          OFFENSE_DESCRIPTION %in% (filtered()  %>% count(OFFENSE_DESCRIPTION) %>% arrange(-n) %>% head(5) %>% pull(OFFENSE_DESCRIPTION))) %>% 
      ggplot(aes(DISTRICT, fill = OFFENSE_DESCRIPTION))+
      geom_bar(position = "fill")
  })
  
  output$mymap <- renderLeaflet({
    
    dm <- data.frame(latitude = filtered()$Lat, longitude = filtered()$Long)
    
    leaflet(dm) %>% addProviderTiles("Stamen.TonerLite") %>%  
      addMarkers(
        clusterOptions = markerClusterOptions()
      ) %>% setView(-71.1, 42.32, zoom = 12)
    
  })
  
  output$table <- renderDataTable(
    filtered(),
    options = list(scrollX = TRUE)
  )
  
}



## Project Boston Crimes: Building Our App
shinyApp(ui, server)