Skip to contents

Data management

Source: vignettes/v03_data_management.Rmd
v03_data_management.Rmd

This vignette1 explains how to prepare or simulate data in fHMM for estimation.

Empirical data

Empirical data can be provided either as a data.frame or as a comma-separated values (.csv) file, see the vignette on specifying the controls for details.2 The fHMM package comes with a dataset of the Deutscher Aktienindex for demonstration purpose that can be accessed as follows:

system.file("extdata", "dax.csv", package = "fHMM")

The prepare_data() function prepares the data based on the data controls specifications and returns an fHMM_data object that can be passed to the fit_model() function for model fitting.

controls <- list(
  states = 3,
  sdds   = "t",
  data   = list(file        = system.file("extdata", "dax.csv", package = "fHMM"),
                date_column = "Date",
                data_column = "Close",
                logreturns  = TRUE)
)
controls <- set_controls(controls)
data <- prepare_data(controls)
summary(data)
#> Summary of fHMM empirical data
#> * number of observations: 9012 
#> * data source: dax.csv 
#> * date column: Date 
#> * log returns: TRUE

Download stock data

Daily stock prices listed on https://finance.yahoo.com/ can be downloaded directly via

download_data(symbol, from, to, file)

where

  • symbol is the stock’s symbol that has to match the official symbol on https://finance.yahoo.com/,3

  • from and to define the time interval (in format "YYYY-MM-DD"),

  • file is the name of the file where the .csv-file is saved. If file = NULL (default), the data is not saved but returned as a data.frame.

For example, the following call downloads the 21st century daily data of the DAX:

dax <- download_data(symbol = "^GDAXI", from = "2000-01-01", to = Sys.Date())
head(dax)
#>         Date    Open    High     Low   Close Adj.Close   Volume
#> 1 2000-01-03 6961.72 7159.33 6720.87 6750.76   6750.76 43072500
#> 2 2000-01-04 6747.24 6755.36 6510.46 6586.95   6586.95 46678400
#> 3 2000-01-05 6585.85 6585.85 6388.91 6502.07   6502.07 52682800
#> 4 2000-01-06 6501.45 6539.31 6402.63 6474.92   6474.92 41180600
#> 5 2000-01-07 6489.94 6791.53 6470.14 6780.96   6780.96 56058900
#> 6 2000-01-10 6785.47 6975.26 6785.47 6925.52   6925.52 42006200

Highlighting events

Historical events can be highlighted by specifying a named list events with elements dates (a vector of dates) and labels (a vector of labels for the events) and passing it to the plot method, for example:

events <- fHMM:::fHMM_events(
  list(
    dates = c("2001-09-11","2008-09-15","2020-01-27"),
    labels = c("9/11 terrorist attack","Bankruptcy of Lehman Brothers","First COVID-19 case in Germany")
    )
  )
print(events)
#>        dates                         labels
#> 1 2001-09-11          9/11 terrorist attack
#> 2 2008-09-15  Bankruptcy of Lehman Brothers
#> 3 2020-01-27 First COVID-19 case in Germany
plot(data, events = events)

Simulated data

If the data parameter in the model’s controls is unspecified, the model is fitted to simulated data from the model specification. This can be useful for testing the functionality or conducting simulation experiments.

True model parameters can be specified by defining an fHMM_parameters-object via the fHMM_parameters() function and passing it to prepare_data().