flex.Rmd

---
# Set up the Environment
title: "DSE Indices"
runtime: shiny
output: 
  flexdashboard::flex_dashboard:
    theme:
      bg: "#fff"
      fg: "#000" 
      primary: "#446E9B"
      secondary: "#446E9B"
      base_font: !expr bslib::font_google("Prompt")
      code_font: !expr bslib::font_google("JetBrains Mono")
      bootswatch: spacelab
    orientation: rows
    vertical_layout: fill
    #social: menu
    source_code: embed
    navbar:
      - { title: "GitHub", href: "https://github.com/tanvird3", align: right, icon: fa-github}
      #- { title: "Home", href: "https://google.com", align: right, icon: fa-facebook }
---

```{r, include = FALSE}
# Import the libraries and define the theme
library(flexdashboard)
library(plotly)
library(readr)
library(dplyr)
library(shiny)
library(shinyjs)
library(prophet)
library(reticulate)
# unquote the following line if the theme customization window needed
#bslib::bs_themer()
thematic::thematic_rmd(
  font = "auto",
  # To get the dark bg on the geom_raster()
  sequential = thematic::sequential_gradient(
    fg_low = FALSE,
    fg_weight = 0,
    bg_weight = 1
  )
)
theme_set(theme_bw(base_size = 20))
```
```{r, include = FALSE}
# load the virtual env
reticulate::virtualenv_create("python35_env", python = "python3")
reticulate::virtualenv_install("python35_env",
                               packages = c("investpy"),
                               ignore_installed = T)
reticulate::use_virtualenv("python35_env", required = TRUE)
```

Sidebar {.sidebar}
=======================================================================
###
```{r}
selectInput(
  inputId = "index",
  label = "Select Index",
  choices =
    c("DSE Broad", "DSE 30", "DSEX Shariah"),
  selected = "DSE Broad"
)

dateRangeInput(
  inputId = "date",
  label = "Choose a Date",
  start = Sys.Date() - 364,
  end = Sys.Date()
)

# The Global ActionButton
actionButton("Generate", "Generate Plots")
```
```{r}
```

```{r}
# Prepare the required tables for plotting
data_gen <- function(startdate, enddate, index) {
  source_python("index.py")
  
  startdate <- format(startdate, "%d/%m/%Y")
  enddate <- format(enddate, "%d/%m/%Y")
  
  df <- index_hist(startdate, enddate, index)
  df <- df %>% mutate(Date = as.Date(Date, format = "%Y-%m-%d"))
  
  # create return series
  df <- df %>% mutate(Return = c((diff(Close) * -1) /Close[-1], 0))
  # Color setting related to values of Return 
  df <- df %>% mutate(Color = if_else(Return >= 0, "#109618", "#D62728"))
  df <- df %>% mutate(WinLose = if_else(Return >= 0, "Winning Days", "Losing Days"))
  
  max_index <- max(df$Close)
  min_index <- min(df$Close)
  avg_index <- mean(df$Close)
  
  pct_var <- (max_index - min_index) / min_index * 100
  pct_change <-
    (df$Close[1] - df$Close[nrow(df)]) / df$Close[nrow(df)] * 100
  
  
  sum_tab <-
    data.frame(
      Particulars = c("Max Index", "Min Index", "Avg Index"),
      Values = c(max_index, min_index, avg_index)
    )
  
  pct_tab <-
    data.frame(
      Particulars = c("Max-Min", "End-Beginning"),
      Values = c(pct_var, pct_change)
    )
  
  winlose_tab <-
    df %>% group_by(WinLose) %>% summarise(Day_Count = n())
  
  # the DSEX Time Plot
  time_plot <- df %>%
    plot_ly(
      x = ~ Date,
      y = ~ Close,
      name = index, 
      type = "scatter",
      mode = "lines",
      fill = "tozeroy",
      showlegend = F
    )
  
  # Return Series Plot
  return_plot <- df %>%
    plot_ly(
      x = ~ Date,
      y = ~ Return,
      name = "Return",
      type = "bar",
      showlegend = F,
      marker = list(color = ~ Color)
    ) 
  
  # combine the plots
  index_plot <-
    subplot(
      time_plot,
      return_plot,
      nrows = 2,
      shareX = TRUE,
      shareY = FALSE,
      which_layout = "merge"
    ) %>%
    layout(
      title = paste("<br>Timeline of", index, "Index"),
      xaxis = list(title = ""),
      xaxis2 = list(title = ""),
      yaxis = list(title = "Index Value"),
      yaxis2 = list(title = "% Return")
    )
  
  # Basic Descriptive Statistics Plot
  summary_plot <- sum_tab %>%
    plot_ly(
      x =  ~ Particulars,
      y =  ~ Values,
      type = "bar",
      name = "Index Movement",
      marker = list(color = "#A777F1")
    ) %>%
    layout(title = paste("<br>", "Movement of", index),
           xaxis = list(title = ""),
           yaxis = list(title = ""))
  
  # Growth Over the Period Plot
  growth_plot <- pct_tab %>%
    plot_ly(
      x =  ~ Particulars,
      y =  ~ Values,
      type = "bar",
      name = "Index Change",
      marker = list(color = "#FF6692")
    ) %>%
    layout(title = paste("<br>", "% Change of", index),
           xaxis = list(title = ""),
           yaxis = list(title = ""))
  
  # Win Lose Day Count Plot
  winlose_plot <-
    winlose_tab %>% plot_ly(
      labels = ~ WinLose,
      values = ~ Day_Count,
      type = "pie",
      showlegend = F
    ) %>%
    layout(title = paste("<br>", index, "Count of W/L Days"))
  
  # Forecast with Prophet Method
  # keep only the required columns
  df_mod <- df %>% select(Date, Close)
  
  # rename the columns as per requirements
  names(df_mod) <- c("ds", "y")
  
  # fit the model
  model_fit <- prophet(df_mod, seasonality.mode = "additive")
  future_get <- make_future_dataframe(model_fit, periods = 64)
  future_get <-
    future_get %>% mutate(wd = weekdays(ds)) %>% filter(!wd %in% c("Friday", "Saturday")) %>% select(-wd)
  forecast_get <-
    predict(model_fit, future_get) %>% select(ds, yhat, yhat_lower, yhat_upper)
  
  # Prepare data fro plotting
  df_dt <- df %>% select(Date, Close)
  names(df_dt)[1] <- "ds"
  df_dt <- df_dt %>% mutate(ds = as.Date(ds, format = "%Y-%m-%d"))
  df_plot <-
    df_dt %>% full_join(forecast_get, by = "ds") %>% arrange(desc(ds)) %>% slice_head(n = 100)
  
  # Plot the forecasts along with original data
  pplot <- plot_ly(df_plot,
                   x = ~ ds) %>%
    
    add_trace(
      y = ~ yhat_upper,
      name = "Upper Band",
      type = "scatter",
      mode = "lines",
      line = list(dash = "dot", color = "#B82E2E"),
      showlegend = F
    ) %>% add_trace(
      y = ~ yhat_lower,
      name = "Lower Band",
      type = "scatter",
      mode = "lines",
      line = list(dash = "dot", color = "#990099"),
      fill = "tonexty",
      fillcolor = "rgba(0,100,80,0.2)",
      showlegend = F
    ) %>%
    
    add_trace(
      y = ~ yhat,
      name = "Forecasted Index",
      type = "scatter",
      mode = "lines",
      line = list(color = "#109618")
    ) %>%
    
    add_trace(
      y = ~ Close,
      type = "scatter",
      mode = "lines",
      name = "Actual Index",
      line = list(color = "blue")
    ) %>%
    
    layout(
      title = paste(index, "Forecast"),
      xaxis = list(title = ""),
      yaxis = list(title = ""),
      margin = list(t = 120),
      legend = list(
        orientation = "h",
        x = 0,
        y = 1.4
      )
    )
  
  # The return series of plots
  return(
    list(
      index_plot = index_plot,
      summary_plot = summary_plot,
      growth_plot = growth_plot,
      winlose_plot = winlose_plot, 
      forecast_plot = pplot
    )
  )
}

# Generate the Output
output_get <-
  eventReactive(input$Generate, {
    data_gen(input$date[1], input$date[2], input$index)
  }, ignoreNULL = F)
```

<!-- # This is the First Tab -->
Analysis
=======================================================================

<!-- # Rows of the Tab -->
Row
-----------------------------------------------------------------------
```{r}
Index_Plot <- renderPlotly(output_get()$index_plot)
Index_Plot
```
```{r}
```
Row
-----------------------------------------------------------------------
```{r}
Summary_Plot <- renderPlotly(output_get()$summary_plot)
Summary_Plot
```
```{r}
Growth_Plot <- renderPlotly(output_get()$growth_plot)
Growth_Plot
```
```{r}
winlose_plot <- renderPlotly(output_get()$winlose_plot)
winlose_plot
```

<!-- # Second Tab -->
Forecast
=======================================================================

```{r}
Forecast_Plot <- renderPlotly(output_get()$forecast_plot)
Forecast_Plot
```