R/get_datas.R

In ataustin/stocking:

#all_stocks <- stockSymbols()

GetStockData <- function(symbol, from = NULL) {
  # Wrapper to get stock data data.frame format -- any value in keeping as xts/zoo?
  # Creates a time-valued date column

  if(is.null(from)) from <- "1900-01-01"
  x <- quantmod::getSymbols(symbol, src = "yahoo", auto.assign = FALSE, from = from)
  x <- as.data.frame(x)

  colnames(x) <- .AdjustStockNames(symbol, names(x))

  x$date <- strptime(rownames(x), format = "%Y-%m-%d", tz = "UTC")

  rownames(x) <- NULL

  out <- list(data   = x,
              symbol = toupper(symbol),
              start  = min(x$date),
              end    = max(x$date))

  structure(out, class = "stock")
}


# StockWindow ----------------------------------------------------------------------------

SubsetStock <- function(stock, start = "min", end = "max") {
  # Returns a window of stock history, as class stock, given by start and end dates

  # Args:
  #   stock: object of class stock
  #   start: first day of stock price to include; defaults to first available in data
  #   end  : last day of stock price to include; defaults to last available in data

  # Returns:
  #   an object of class stock

  # Supply defaults
  if(start == "min" || start < stock$data$date[1L]) {
    start <- stock$data$date[1L]
  }

  if(end == "max" || end > stock$data$date[nrow(stock$data)]) {
    end <- stock$data$date[nrow(stock$data)]
  }

  # Subset data and replace data with subsetted version
  stock_data <- stock$data[stock$data$date >= start &
                             stock$data$date <= end, ]
  stock$data  <- stock_data
  stock$start <- stock$data$date[1L]
  stock$end   <- stock$data$date[nrow(stock$data)]

  # Return output
  return(stock)
}


SplitStock <- function(stock, split, start = "min", end = "max") {
  # Splits stock data into two objects, one for training and one for testing.

  # Args:
  #   stock: object of class "stock"
  #   split: the DATE on which to begin the test data in yyyy-mm-dd format
  #   start: optional start date of TRAINING data.  If not specified,
  #          the earliest date ("min") in the pair object will be used.
  #   end:   the latest data of the TEST data.  If not specified,
  #          the latest date ("max") in the pair object will be used.

  # Control
  if(!is.stock(stock)) {
    stop("Please provide an object of class 'stock'.")
  }

  split <- if(split %in% stock$data$date) {
    train_end <- max(stock$data$date[stock$data$date < split])
  } else {
    train_end <- split
  }

  train <- SubsetStock(stock, start = start, end = train_end)
  # if(train$end == split) {
  #   train <- SubsetStock(train, end = train$data$date[length(train$data$date) - 1])
  # }

  test  <- SubsetStock(stock, start = split, end = end)

  out <- list(train = train,
              test  = test)

  return(out)
}


#######################################################################

is.stock <- function(x) {
  inherits(x, "stock")
}


#######################################################################

print.stock <- function(x, ...) {
  print(head(x$data, 5L))
  cat("  ...\n")
  print(tail(x$data, 5L))
}


#######################################################################

plot.stock <- function(x, ...) {
  # Simple plotting utility for class stock

  # Args:
  #   x: object of class stock

  prepped_data <- .PrepStockForPlot(x$data)
  title <- paste0(x$symbol, ": Open and Close Prices with High/Low Band")

  ggplot(prepped_data, aes(x = date, y = price)) +
    geom_ribbon(aes(ymin = low, ymax = high), fill = "grey70", alpha = 0.3) +
    geom_line(aes(color = time), lwd = 1.3) +
    scale_color_viridis(discrete = TRUE, option = "plasma", name = "Time") +
    labs(x = "Date", y = "Price", title = title) +
    theme_bw()
}


#######################################################################

.PrepStockForPlot <- function(stock_data) {
  # Prepares stock data from object of class "stock" for plotting with ggplot2

  price <- c(stock_data$open, stock_data$close)
  time  <- factor(rep(c("Open", "Close"), each = nrow(stock_data)), levels = c("Open", "Close"))
  date  <- rep(stock_data$date, times = 2L)
  high  <- rep(stock_data$high, times = 2L)
  low   <- rep(stock_data$low, times = 2L)

  plot_data <- data.frame(date = date,
                          price = price,
                          time = time,
                          high = high,
                          low = low)

  return(plot_data)
}


#######################################################################

.MakeFirstLetterLowerCase <- function(char) {
  # Makes first letter of vector char into lower case

  first_letters <- substr(char, 1L, 1L)
  lower_firsts  <- tolower(first_letters)
  char_ending   <- substr(char, 2L, nchar(char))

  paste0(lower_firsts, char_ending)
}


#######################################################################

.AdjustStockNames <- function(symbol, data_names) {
  # Standardizes names of stock data coming from quantmod::getSymbols
  # Returns character vector of adjusted names

  gsub_pattern   <- paste0(toupper(symbol), "\\.")
  char_no_symbol <- gsub(gsub_pattern, "", data_names)

  char_out <- .MakeFirstLetterLowerCase(char_no_symbol)

  return(char_out)
}