Introduction for the TSstudio Package

Overview

The TSstudio (time series studio) package is a toolbox for interactive visualization applications for time series analysis with the Plotly package. The package is under development and first iteration is available on github.

Installation

The package is available for installation with the devtools package (if devetools package is not installed please use install.packages("devtools") to install it).

# Install the TSstudio
devtools::install_github("RamiKrispin/TSstudio")

Key Features

Visualization of fitted and forecasted values vs. actual values

library(TSstudio)
library(forecast)

# Load the AirPassengers dataset
data("AirPassengers")

# set the forecast horizon for 12 months
h <- 12 

# Split the data into training and testing sets (leaving the last 12 months for testing)
train <- window(AirPassengers, 
                start = time(AirPassengers)[1], 
                end = time(AirPassengers)[length(AirPassengers) - h])
test <- window(AirPassengers, 
               start = time(AirPassengers)[length(AirPassengers) - h + 1], 
               end = time(AirPassengers)[length(AirPassengers)])

# Using auto.arima to train and forecast the last 12 months on the series
fc <- forecast(auto.arima(train, lambda = BoxCox.lambda(train)), h = h)

# Plotting the series vs the fitted and the forecasted
fortest_ly(actual = AirPassengers, forecast.obj = fc, train = train, test = test)

Seasonality plot

seasonal_ly(AirPassengers)

Plotting time series object

# Simple plot
ts.plot_ly(AirPassengers)

# Adding slider and markers, changing the color to green
ts.plot_ly(AirPassengers, 
           slider = TRUE, 
           line.mode = "lines+markers", 
           color = "green", 
           width = 2)

Plotting the acf and the pacf estimation

#plotting the acf and the pacf estimation
# setting the lags to 60 and the level of significant to 0.01
acf_ly(AirPassengers, lag.max = 60, ci = 0.99)

pacf_ly(AirPassengers, lag.max = 60, ci = 0.99)