| Title: | Functions for Time Series Analysis and Forecasting |
|---|---|
| Description: | Provides a set of tools for descriptive and predictive analysis of time series data. That includes functions for interactive visualization of time series objects and as well utility functions for automation time series forecasting. |
| Authors: | Rami Krispin [aut, cre] |
| Maintainer: | Rami Krispin <[email protected]> |
| License: | MIT + file LICENSE |
| Version: | 0.1.7 |
| Built: | 2024-11-01 03:40:28 UTC |
| Source: | https://github.com/ramikrispin/tsstudio |
Diagnostic Plots for ARIMA Models
arima_diag(ts.obj, method = list(first = list(diff = 1, log = TRUE, title = "First Difference with Log Transformation")), cor = TRUE)arima_diag(ts.obj, method = list(first = list(diff = 1, log = TRUE, title = "First Difference with Log Transformation")), cor = TRUE)
ts.obj |
A ts object |
method |
A list, defines the transformation parameters of each plot. Each plot should be defined by a list, where the name of the list defines the plot ID. The plot parameters are: diff - an integer, defines the degree of diffrence log - a boolean, optional, defines if log transformation should be used title - optional, the plot title |
cor |
A boolean, if TRUE (default), will plot the series ACF and PACF |
The arima_diag function provides a set of diagnostic plots for identify the ARIMA model parameters. The ACF and PACF can assist in identifying the AR and MA process, and the diffrence plotting hel in idenitfying the degree of differencing that required to make the series stationary
A plot
data(USgas) arima_diag(ts.obj = USgas) # Can define more than one differencing plot using the 'method' argument arima_diag(ts.obj = USgas, cor = TRUE, method = list(first = list(diff = 1, log = TRUE, title = "First Diff with Log Transformation"), Second = list(diff = c(1,1), log = TRUE, title = "Second Diff with Log Transformation")))data(USgas) arima_diag(ts.obj = USgas) # Can define more than one differencing plot using the 'method' argument arima_diag(ts.obj = USgas, cor = TRUE, method = list(first = list(diff = 1, log = TRUE, title = "First Diff with Log Transformation"), Second = list(diff = c(1,1), log = TRUE, title = "Second Diff with Log Transformation")))
Visualize the series y against the series x lags (according to the setting of the lags argument) and return the corresponding cross-correlation value for each lag
ccf_plot(x, y, lags = 0:12, margin = 0.02, n_plots = 3, Xshare = TRUE, Yshare = TRUE, title = NULL)ccf_plot(x, y, lags = 0:12, margin = 0.02, n_plots = 3, Xshare = TRUE, Yshare = TRUE, title = NULL)
x |
A univariate time series object of a class "ts" |
y |
A univariate time series object of a class "ts" |
lags |
An integer, set the lags range, by default will plot the two series along with the first 12 lags |
margin |
Plotly parameter, either a single value or four values (all between 0 and 1). If four values provided, the first will be used as the left margin, the second will be used as the right margin, the third will be used as the top margin, and the fourth will be used as the bottom margin. If a single value provided, it will be used as all four margins. |
n_plots |
An integer, define the number of plots per row |
Xshare |
Plotly parameter, should the x-axis be shared amongst the subplots? |
Yshare |
Plotly parameter, should the y-axis be shared amongst the subplots? |
title |
A character, optional, set the plot title |
Plot
data("USUnRate") data("USVSales") ccf_plot(x = USVSales, y = USUnRate) #Plotting the first 6 lead and lags of the USVSales with the USUnRate ccf_plot(x = USVSales, y = USUnRate, lags = -6:6) # Setting the plot margin and number of plots in each raw ccf_plot(x = USVSales, y = USUnRate, lags = c(0, 6, 12, 24), margin = 0.01, n_plots = 2)data("USUnRate") data("USVSales") ccf_plot(x = USVSales, y = USUnRate) #Plotting the first 6 lead and lags of the USVSales with the USUnRate ccf_plot(x = USVSales, y = USUnRate, lags = -6:6) # Setting the plot margin and number of plots in each raw ccf_plot(x = USVSales, y = USUnRate, lags = c(0, 6, 12, 24), margin = 0.01, n_plots = 2)
Provides a visualization of the residuals of a time series model. That includes a time series plot of the residuals, and the plots of the autocorrelation function (acf) and histogram of the residuals
check_res(ts.model, lag.max = 36)check_res(ts.model, lag.max = 36)
ts.model |
A time series model (or forecasted) object, support any model from the forecast package with a residuals output |
lag.max |
The maximum number of lags to display in the residuals' autocorrelation function plot |
library(forecast) data(USgas) # Create a model fit <- auto.arima(USgas) # Check the residuals of the model check_res(fit)library(forecast) data(USgas) # Create a model fit <- auto.arima(USgas) # Check the residuals of the model check_res(fit)
Coffee Prices: Robusta and Arabica: 1960 - 2018. Units: Dollars per Kg
Coffee_PricesCoffee_Prices
Time series data - 'mts' object
WIKI Commodity Prices - Quandle
ts_plot(Coffee_Prices)ts_plot(Coffee_Prices)
train_model ComponentsAdd, edit, or remove the components of the train_model function
create_model() add_input(model.obj, input) add_methods(model.obj, methods) remove_methods(model.obj, method_ids) add_train_method(model.obj, train_method) add_horizon(model.obj, horizon) build_model(model.obj) set_error(model.obj, error) add_xreg(model.obj, xreg) add_level(model.obj, level)create_model() add_input(model.obj, input) add_methods(model.obj, methods) remove_methods(model.obj, method_ids) add_train_method(model.obj, train_method) add_horizon(model.obj, horizon) build_model(model.obj) set_error(model.obj, error) add_xreg(model.obj, xreg) add_level(model.obj, level)
model.obj |
The train_model skeleton, created by the create_model function or edited by add_input, add_methods, remove_methods, add_train_method or add_horizon |
input |
A univariate time series object (ts class) |
methods |
A list, defines the models to use for training and forecasting the series. The list must include a sub list with the model type, and the model's arguments (when applicable) and notes about the model. The sub-list name will be used as the model ID. Possible models:
|
method_ids |
A character, defines the IDs of the model methods to be remove with the remove_methods function |
train_method |
A list, defines the train approach, either using a single testing partition (sample out) or use multiple testing partitions (backtesting). The list should include the training method argument, (please see 'details' for the structure of the argument) |
horizon |
An integer, defines the forecast horizon |
error |
A character, defines the error metrics to be used to sort the models leaderboard. Possible metric - "MAPE" or "RMSE" |
xreg |
Optional, a list with two vectors (e.g., data.frame or matrix) of external regressors, one vector corresponding to the input series and second to the forecast itself (e.g., must have the same length as the input and forecast horizon, respectively) |
level |
An integer, set the confidence level of the prediction intervals |
## Not run: ### Building train_model function by adding its different components # Create a skeleton model md <- create_model() class(md) # Add input data(USgas) md <- add_input(model.obj = md, input = USgas) # Add methods methods <- list(ets1 = list(method = "ets", method_arg = list(opt.crit = "lik"), notes = "ETS model with opt.crit = lik"), ets2 = list(method = "ets", method_arg = list(opt.crit = "amse"), notes = "ETS model with opt.crit = amse"), arima1 = list(method = "arima", method_arg = list(order = c(1,1,1), seasonal = list(order = c(1,0,1))), notes = "SARIMA(1,1,1)(1,0,1)")) md <- add_methods(model.obj = md, methods = methods) # Add additional methods methods2 <- list(arima2 = list(method = "arima", method_arg = list(order = c(2,1,2), seasonal = list(order = c(1,1,1))), notes = "SARIMA(2,1,2)(1,1,1)"), hw = list(method = "HoltWinters", method_arg = NULL, notes = "HoltWinters Model"), tslm = list(method = "tslm", method_arg = list(formula = input ~ trend + season), notes = "tslm model with trend and seasonal components")) md <- add_methods(model.obj = md, methods = methods2) # Remove methods md <- remove_methods(model.obj = md, method_ids = c("ets2")) # Add train method md <- add_train_method(model.obj = md, train_method = list(partitions = 6, sample.out = 12, space = 3)) # Set the forecast horizon md <- add_horizon(model.obj = md, horizon = 12) # Add the forecast prediction intervals confidence level md <- add_level(model.obj = md, level = c(90, 95)) ### Alternatively, pipe the function with the magrittr package library(magrittr) md <- create_model() %>% add_input(input = USgas) %>% add_methods(methods = methods) %>% add_methods(methods = methods2) %>% add_train_method(train_method = list(partitions = 4, sample.out = 12, space = 3)) %>% add_horizon(horizon = 12) %>% add_level(level = c(90, 95)) # Run the model fc <- md %>% build_model() ## End(Not run)## Not run: ### Building train_model function by adding its different components # Create a skeleton model md <- create_model() class(md) # Add input data(USgas) md <- add_input(model.obj = md, input = USgas) # Add methods methods <- list(ets1 = list(method = "ets", method_arg = list(opt.crit = "lik"), notes = "ETS model with opt.crit = lik"), ets2 = list(method = "ets", method_arg = list(opt.crit = "amse"), notes = "ETS model with opt.crit = amse"), arima1 = list(method = "arima", method_arg = list(order = c(1,1,1), seasonal = list(order = c(1,0,1))), notes = "SARIMA(1,1,1)(1,0,1)")) md <- add_methods(model.obj = md, methods = methods) # Add additional methods methods2 <- list(arima2 = list(method = "arima", method_arg = list(order = c(2,1,2), seasonal = list(order = c(1,1,1))), notes = "SARIMA(2,1,2)(1,1,1)"), hw = list(method = "HoltWinters", method_arg = NULL, notes = "HoltWinters Model"), tslm = list(method = "tslm", method_arg = list(formula = input ~ trend + season), notes = "tslm model with trend and seasonal components")) md <- add_methods(model.obj = md, methods = methods2) # Remove methods md <- remove_methods(model.obj = md, method_ids = c("ets2")) # Add train method md <- add_train_method(model.obj = md, train_method = list(partitions = 6, sample.out = 12, space = 3)) # Set the forecast horizon md <- add_horizon(model.obj = md, horizon = 12) # Add the forecast prediction intervals confidence level md <- add_level(model.obj = md, level = c(90, 95)) ### Alternatively, pipe the function with the magrittr package library(magrittr) md <- create_model() %>% add_input(input = USgas) %>% add_methods(methods = methods) %>% add_methods(methods = methods2) %>% add_train_method(train_method = list(partitions = 4, sample.out = 12, space = 3)) %>% add_horizon(horizon = 12) %>% add_level(level = c(90, 95)) # Run the model fc <- md %>% build_model() ## End(Not run)
Crude Oil Prices: Brent - Europe: 1987 - 2019. Units: Dollars per Barrel
EURO_BrentEURO_Brent
Time series data - 'zoo' object
U.S. Energy Information Administration, Crude Oil Prices: Brent - Europe [MCOILBRENTEU], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/MCOILBRENTEU, January 8, 2018.
ts_plot(EURO_Brent) ts_decompose(EURO_Brent, type = "both")ts_plot(EURO_Brent) ts_decompose(EURO_Brent, type = "both")
Creating different forecast paths for forecast objects (when applicable),
by utilizing the underline model distribution with the simulate function
forecast_sim(model, h, n, sim_color = "blue", opacity = 0.05, plot = TRUE)forecast_sim(model, h, n, sim_color = "blue", opacity = 0.05, plot = TRUE)
model |
A forecasting model supporting |
h |
An integer, defines the forecast horizon |
n |
An integer, set the number of iterations of the simulation |
sim_color |
Set the color of the simulation paths lines |
opacity |
Set the opacity level of the simulation path lines |
plot |
Logical, if TRUE will desplay the output plot |
The baseline series, the simulated values and a plot
## Not run: library(forecast) data(USgas) # Create a model fit <- auto.arima(USgas) # Simulate 100 possible forecast path, with horizon of 60 months forecast_sim(model = fit, h = 60, n = 100) ## End(Not run)## Not run: library(forecast) data(USgas) # Create a model fit <- auto.arima(USgas) # Simulate 100 possible forecast path, with horizon of 60 months forecast_sim(model = fit, h = 60, n = 100) ## End(Not run)
University of Michigan Consumer Survey, Index of Consumer Sentiment: 1980 - 2019. Units: Index 1966:Q1=100
Michigan_CSMichigan_CS
Time series data - 'xts' object
University of Michigan, University of Michigan: Consumer Sentiment
ts_plot(Michigan_CS) ts_heatmap(Michigan_CS)ts_plot(Michigan_CS) ts_heatmap(Michigan_CS)
Plot the Models Error Rates on the Testing Partitions
plot_error(model.obj, error = "MAPE", palette = "Set1")plot_error(model.obj, error = "MAPE", palette = "Set1")
model.obj |
A train_model object |
error |
A character, defines the type of error metrics to plot, possible metric - "MAPE" or "RMSE" |
palette |
A character, defines the color type to used on the plot, use row.names(RColorBrewer::brewer.pal.info) to view possible color palletes |
The plot_model provides a visualization of the models performance on the testing paritions for the train_model function output
A plot with a summery of the models error rate by testing partition
## Not run: # Defining the models and their arguments methods <- list(ets1 = list(method = "ets", method_arg = list(opt.crit = "lik"), notes = "ETS model with opt.crit = lik"), ets2 = list(method = "ets", method_arg = list(opt.crit = "amse"), notes = "ETS model with opt.crit = amse"), arima1 = list(method = "arima", method_arg = list(order = c(2,1,0)), notes = "ARIMA(2,1,0)"), arima2 = list(method = "arima", method_arg = list(order = c(2,1,2), seasonal = list(order = c(1,1,1))), notes = "SARIMA(2,1,2)(1,1,1)"), hw = list(method = "HoltWinters", method_arg = NULL, notes = "HoltWinters Model"), tslm = list(method = "tslm", method_arg = list(formula = input ~ trend + season), notes = "tslm model with trend and seasonal components")) # Training the models with backtesting md <- train_model(input = USgas, methods = methods, train_method = list(partitions = 6, sample.out = 12, space = 3), horizon = 12, error = "MAPE") # Plot the models performance on the testing partitions plot_error(model.obj = md) ## End(Not run)## Not run: # Defining the models and their arguments methods <- list(ets1 = list(method = "ets", method_arg = list(opt.crit = "lik"), notes = "ETS model with opt.crit = lik"), ets2 = list(method = "ets", method_arg = list(opt.crit = "amse"), notes = "ETS model with opt.crit = amse"), arima1 = list(method = "arima", method_arg = list(order = c(2,1,0)), notes = "ARIMA(2,1,0)"), arima2 = list(method = "arima", method_arg = list(order = c(2,1,2), seasonal = list(order = c(1,1,1))), notes = "SARIMA(2,1,2)(1,1,1)"), hw = list(method = "HoltWinters", method_arg = NULL, notes = "HoltWinters Model"), tslm = list(method = "tslm", method_arg = list(formula = input ~ trend + season), notes = "tslm model with trend and seasonal components")) # Training the models with backtesting md <- train_model(input = USgas, methods = methods, train_method = list(partitions = 6, sample.out = 12, space = 3), horizon = 12, error = "MAPE") # Plot the models performance on the testing partitions plot_error(model.obj = md) ## End(Not run)
Visualization functions for forecast package forecasting objects
plot_forecast(forecast_obj, title = NULL, Xtitle = NULL, Ytitle = NULL, color = NULL, width = 2)plot_forecast(forecast_obj, title = NULL, Xtitle = NULL, Ytitle = NULL, color = NULL, width = 2)
forecast_obj |
A forecast object from the forecast, forecastHybrid, or bsts packages |
title |
A character, a plot title, optional |
Xtitle |
Set the X axis title, default set to NULL |
Ytitle |
Set the Y axis title, default set to NULL |
color |
A character, the plot, support both name and expression |
width |
An Integer, define the plot width, default is set to 2 |
data(USgas) library(forecast) fit <- ets(USgas) fc<- forecast(fit, h = 60) plot_forecast(fc)data(USgas) library(forecast) fit <- ets(USgas) fc<- forecast(fit, h = 60) plot_forecast(fc)
Visualizing Grid Search Results
plot_grid(grid.obj, top = NULL, highlight = 0.1, type = "parcoords", colors = list(showscale = TRUE, reversescale = FALSE, colorscale = "Jet"))plot_grid(grid.obj, top = NULL, highlight = 0.1, type = "parcoords", colors = list(showscale = TRUE, reversescale = FALSE, colorscale = "Jet"))
grid.obj |
A ts_grid output object |
top |
An integer, set the number of hyper-parameters combinations to visualize (ordered by accuracy). If set to NULL (default), will plot the top 100 combinations |
highlight |
A proportion between 0 (excluding) and 1, set the number of hyper-parameters combinations to highlight (by accuracy), if the type argument is set to "parcoords" |
type |
The plot type, either "3D" for 3D plot or "parcoords" for parallel coordinates plot. Note: the 3D plot option is applicable whenever there are three tuning parameters, otherwise will use a 2D plot for two tuning parameters. |
colors |
A list of plotly arguments for the color scale setting: showscale - display the color scale if set to TRUE. reversescale - reverse the color scale if set to TRUE colorscale set the color scale of the plot, possible palettes are: Greys, YlGnBu, Greens , YlOrRd, Bluered, RdBu, Reds, Blues, Picnic, Rainbow, Portland, Jet, Hot, Blackbody, Earth, Electric, Viridis, Cividis |
Plot the Models Performance on the Testing Partitions
plot_model(model.obj, model_ids = NULL)plot_model(model.obj, model_ids = NULL)
model.obj |
A train_model object |
model_ids |
A character, defines the trained models to plot, if set to NULL (default), will plot all the models |
The plot_model provides a visualization of the models performance on the testing paritions for the train_model function output
Animation of models forecast on the testing partitions compared to the actuals
## Not run: # Defining the models and their arguments methods <- list(ets1 = list(method = "ets", method_arg = list(opt.crit = "lik"), notes = "ETS model with opt.crit = lik"), ets2 = list(method = "ets", method_arg = list(opt.crit = "amse"), notes = "ETS model with opt.crit = amse"), arima1 = list(method = "arima", method_arg = list(order = c(2,1,0)), notes = "ARIMA(2,1,0)"), arima2 = list(method = "arima", method_arg = list(order = c(2,1,2), seasonal = list(order = c(1,1,1))), notes = "SARIMA(2,1,2)(1,1,1)"), hw = list(method = "HoltWinters", method_arg = NULL, notes = "HoltWinters Model"), tslm = list(method = "tslm", method_arg = list(formula = input ~ trend + season), notes = "tslm model with trend and seasonal components")) # Training the models with backtesting md <- train_model(input = USgas, methods = methods, train_method = list(partitions = 6, sample.out = 12, space = 3), horizon = 12, error = "MAPE") # Plot the models performance on the testing partitions plot_model(model.obj = md) # Plot only the ETS models plot_model(model.obj = md , model_ids = c("ets1", "ets2")) ## End(Not run)## Not run: # Defining the models and their arguments methods <- list(ets1 = list(method = "ets", method_arg = list(opt.crit = "lik"), notes = "ETS model with opt.crit = lik"), ets2 = list(method = "ets", method_arg = list(opt.crit = "amse"), notes = "ETS model with opt.crit = amse"), arima1 = list(method = "arima", method_arg = list(order = c(2,1,0)), notes = "ARIMA(2,1,0)"), arima2 = list(method = "arima", method_arg = list(order = c(2,1,2), seasonal = list(order = c(1,1,1))), notes = "SARIMA(2,1,2)(1,1,1)"), hw = list(method = "HoltWinters", method_arg = NULL, notes = "HoltWinters Model"), tslm = list(method = "tslm", method_arg = list(formula = input ~ trend + season), notes = "tslm model with trend and seasonal components")) # Training the models with backtesting md <- train_model(input = USgas, methods = methods, train_method = list(partitions = 6, sample.out = 12, space = 3), horizon = 12, error = "MAPE") # Plot the models performance on the testing partitions plot_model(model.obj = md) # Plot only the ETS models plot_model(model.obj = md , model_ids = c("ets1", "ets2")) ## End(Not run)
Histogram plot of the residuals values
res_hist(forecast.obj)res_hist(forecast.obj)
forecast.obj |
A fitted or forecasted object (of the forecast package) with residuals output |
## Not run: library(forecast) data(USgas) # Set the horizon of the forecast h <- 12 # split to training/testing partition split_ts <- ts_split(USgas, sample.out = h) train <- split_ts$train test <- split_ts$test # Create forecast object fc <- forecast(auto.arima(train, lambda = BoxCox.lambda(train)), h = h) # Plot the fitted and forecasted vs the actual values res_hist(forecast.obj = fc) ## End(Not run)## Not run: library(forecast) data(USgas) # Set the horizon of the forecast h <- 12 # split to training/testing partition split_ts <- ts_split(USgas, sample.out = h) train <- split_ts$train test <- split_ts$test # Create forecast object fc <- forecast(auto.arima(train, lambda = BoxCox.lambda(train)), h = h) # Plot the fitted and forecasted vs the actual values res_hist(forecast.obj = fc) ## End(Not run)
Visualize the fitted values of the training set and the forecast values of the testing set against the actual values of the series
test_forecast(actual, forecast.obj, train = NULL, test, Ygrid = FALSE, Xgrid = FALSE, hover = TRUE)test_forecast(actual, forecast.obj, train = NULL, test, Ygrid = FALSE, Xgrid = FALSE, hover = TRUE)
actual |
The full time series object (supports "ts", "zoo" and "xts" formats) |
forecast.obj |
The forecast output of the training set with horizon align to the length of the testing (support forecasted objects from the “forecast” package) |
train |
Training partition, a subset of the first n observation in the series (not requiredthed) |
test |
The testing (hold-out) partition |
Ygrid |
Logic,show the Y axis grid if set to TRUE |
Xgrid |
Logic,show the X axis grid if set to TRUE |
hover |
If TRUE add tooltip with information about the model accuracy |
## Not run: library(forecast) data(USgas) # Set the horizon of the forecast h <- 12 # split to training/testing partition split_ts <- ts_split(USgas, sample.out = h) train <- split_ts$train test <- split_ts$test # Create forecast object fc <- forecast(auto.arima(train, lambda = BoxCox.lambda(train)), h = h) # Plot the fitted and forecasted vs the actual values test_forecast(actual = USgas, forecast.obj = fc, test = test) ## End(Not run)## Not run: library(forecast) data(USgas) # Set the horizon of the forecast h <- 12 # split to training/testing partition split_ts <- ts_split(USgas, sample.out = h) train <- split_ts$train test <- split_ts$test # Create forecast object fc <- forecast(auto.arima(train, lambda = BoxCox.lambda(train)), h = h) # Plot the fitted and forecasted vs the actual values test_forecast(actual = USgas, forecast.obj = fc, test = test) ## End(Not run)
Method for train test and compare multiple time series models using either one partition (i.e., sample out) or multipe partitions (backtesting)
train_model(input, methods, train_method, horizon, error = "MAPE", xreg = NULL, level = c(80, 95))train_model(input, methods, train_method, horizon, error = "MAPE", xreg = NULL, level = c(80, 95))
input |
A univariate time series object (ts class) |
methods |
A list, defines the models to use for training and forecasting the series. The list must include a sub list with the model type, and the model's arguments (when applicable) and notes about the model. The sub-list name will be used as the model ID. Possible models:
|
train_method |
A list, defines the backtesting parameters: partitions - an integer, set the number of training and testing partitions to be used in the backtesting process, where when partition is set to 1 it is a simple holdout training approach space - an integer, defines the length of the backtesting window expansion sample.in - an integer, optional, defines the length of the training partitions, and therefore the backtesting window structure. By default, it set to NULL and therefore, the backtesting using expending window. Otherwise, when the sample.in defined, the window structure is sliding sample.in - an integer, optional, defines the length of the training partitions, and therefore the type of the backtesting window. By default, is set to NULL, which implay that the backtesting is using an expending window. Otherwise, when defining the size of the training partition, th defines the train approach, either using a single testing partition (sample out) or use multiple testing partitions (backtesting). The list should include the training method argument, (please see 'details' for the structure of the argument) |
horizon |
An integer, defines the forecast horizon |
error |
A character, defines the error metrics to be used to sort the models leaderboard. Possible metric - "MAPE" or "RMSE" |
xreg |
Optional, a list with two vectors (e.g., data.frame or matrix) of external regressors, one vector corresponding to the input series and second to the forecast itself (e.g., must have the same length as the input and forecast horizon, respectively) |
level |
An integer, set the confidence level of the prediction intervals |
## Not run: # Defining the models and their arguments methods <- list(ets1 = list(method = "ets", method_arg = list(opt.crit = "lik"), notes = "ETS model with opt.crit = lik"), ets2 = list(method = "ets", method_arg = list(opt.crit = "amse"), notes = "ETS model with opt.crit = amse"), arima1 = list(method = "arima", method_arg = list(order = c(2,1,0)), notes = "ARIMA(2,1,0)"), arima2 = list(method = "arima", method_arg = list(order = c(2,1,2), seasonal = list(order = c(1,1,1))), notes = "SARIMA(2,1,2)(1,1,1)"), hw = list(method = "HoltWinters", method_arg = NULL, notes = "HoltWinters Model"), tslm = list(method = "tslm", method_arg = list(formula = input ~ trend + season), notes = "tslm model with trend and seasonal components")) # Training the models with backtesting md <- train_model(input = USgas, methods = methods, train_method = list(partitions = 4, sample.out = 12, space = 3), horizon = 12, error = "MAPE") # View the model performance on the backtesting partitions md$leaderboard ## End(Not run)## Not run: # Defining the models and their arguments methods <- list(ets1 = list(method = "ets", method_arg = list(opt.crit = "lik"), notes = "ETS model with opt.crit = lik"), ets2 = list(method = "ets", method_arg = list(opt.crit = "amse"), notes = "ETS model with opt.crit = amse"), arima1 = list(method = "arima", method_arg = list(order = c(2,1,0)), notes = "ARIMA(2,1,0)"), arima2 = list(method = "arima", method_arg = list(order = c(2,1,2), seasonal = list(order = c(1,1,1))), notes = "SARIMA(2,1,2)(1,1,1)"), hw = list(method = "HoltWinters", method_arg = NULL, notes = "HoltWinters Model"), tslm = list(method = "tslm", method_arg = list(formula = input ~ trend + season), notes = "tslm model with trend and seasonal components")) # Training the models with backtesting md <- train_model(input = USgas, methods = methods, train_method = list(partitions = 4, sample.out = 12, space = 3), horizon = 12, error = "MAPE") # View the model performance on the backtesting partitions md$leaderboard ## End(Not run)
An Interactive Visualization of the ACF and PACF Functions
ts_cor(ts.obj, type = "both", seasonal = TRUE, ci = 0.95, lag.max = NULL, seasonal_lags = NULL)ts_cor(ts.obj, type = "both", seasonal = TRUE, ci = 0.95, lag.max = NULL, seasonal_lags = NULL)
ts.obj |
A univariate time series object class 'ts' |
type |
A character, defines the plot type - 'acf' for ACF plot, 'pacf' for PACF plot, and 'both' (default) for both ACF and PACF plots |
seasonal |
A boolean, when set to TRUE (default) will color the seasonal lags |
ci |
The significant level of the estimation - a numeric value between 0 and 1, default is set for 0.95 |
lag.max |
maximum lag at which to calculate the acf. Default is 10*log10(N/m) where N is the number of observations and m the number of series. Will be automatically limited to one less than the number of observations in the series |
seasonal_lags |
A vector of integers, highlight specific cyclic lags (besides the main seasonal lags of the series). This is useful when working with multiseasonal time series data. For example, for a monthly series (e.g., frequency 12) setting the argument to 3 will highlight the quarterly lags |
data(USgas) ts_cor(ts.obj = USgas) # Setting the maximum number of lags to 72 ts_cor(ts.obj = USgas, lag.max = 72) # Plotting only ACF ts_cor(ts.obj = USgas, lag.max = 72, type = "acf")data(USgas) ts_cor(ts.obj = USgas) # Setting the maximum number of lags to 72 ts_cor(ts.obj = USgas, lag.max = 72) # Plotting only ACF ts_cor(ts.obj = USgas, lag.max = 72, type = "acf")
Interactive visualization the trend, seasonal and random components of a time series based on the decompose function from the stats package.
ts_decompose(ts.obj, type = "additive", showline = TRUE)ts_decompose(ts.obj, type = "additive", showline = TRUE)
ts.obj |
a univariate time series object of a class "ts", "zoo" or "xts" |
type |
Set the type of the seasonal component, can be set to either "additive", "multiplicative" or "both" to compare between the first two options (default set to “additive”) |
showline |
Logic, add a separation line between each of the plot components (default set to TRUE) |
# Defualt decompose plot ts_decompose(AirPassengers) # Remove the sepration lines between the plot components ts_decompose(AirPassengers, showline = FALSE) # Plot side by side a decompose of additive and multiplicative series ts_decompose(AirPassengers, type = "both")# Defualt decompose plot ts_decompose(AirPassengers) # Remove the sepration lines between the plot components ts_decompose(AirPassengers, showline = FALSE) # Plot side by side a decompose of additive and multiplicative series ts_decompose(AirPassengers, type = "both")
Tuning time series models with grid search approach using backtesting method. If set to "auto" (default), will use all available cores in the system minus 1
ts_grid(ts.obj, model, optim = "MAPE", periods, window_length = NULL, window_space, window_test, hyper_params, parallel = TRUE, n.cores = "auto")ts_grid(ts.obj, model, optim = "MAPE", periods, window_length = NULL, window_space, window_test, hyper_params, parallel = TRUE, n.cores = "auto")
ts.obj |
A univariate time series object of a class "ts" |
model |
A string, defines the model c("HoltWinters"), currently support only Holt-Winters model |
optim |
A string, set the optimization method - c("MAPE", "RMSE") |
periods |
A string, set the number backtesting periods |
window_length |
An integer, defines the length of the backtesting training window. If set to NULL (default) will use an expending window starting the from the first observation, otherwise will use a sliding window. |
window_space |
An integer, set the space length between each of the backtesting training partition |
window_test |
An integer, set the length of the backtesting testing partition |
hyper_params |
A list, defines the tuning parameters and their range |
parallel |
Logical, if TRUE use multiple cores in parallel |
n.cores |
Set the number of cores to use if the parallel argument is set to TRUE. If set to "auto" (default), will use n-1 of the available cores |
A list
## Not run: data(USgas) # Starting with a shallow search (sequence between 0 and 1 with jumps of 0.1) # To speed up the process, will set the parallel option to TRUE # to run the search in parallel using 8 cores hw_grid_shallow <- ts_grid(ts.obj = USgas, periods = 6, model = "HoltWinters", optim = "MAPE", window_space = 6, window_test = 12, hyper_params = list(alpha = seq(0.01, 1,0.1), beta = seq(0.01, 1,0.1), gamma = seq(0.01, 1,0.1)), parallel = TRUE, n.cores = 8) # Use the parameter range of the top 20 models # to set a narrow but more agressive search a_min <- min(hw_grid_shallow$grid_df$alpha[1:20]) a_max <- max(hw_grid_shallow$grid_df$alpha[1:20]) b_min <- min(hw_grid_shallow$grid_df$beta[1:20]) b_max <- max(hw_grid_shallow$grid_df$beta[1:20]) g_min <- min(hw_grid_shallow$grid_df$gamma[1:20]) g_max <- max(hw_grid_shallow$grid_df$gamma[1:20]) hw_grid_second <- ts_grid(ts.obj = USgas, periods = 6, model = "HoltWinters", optim = "MAPE", window_space = 6, window_test = 12, hyper_params = list(alpha = seq(a_min, a_max,0.05), beta = seq(b_min, b_max,0.05), gamma = seq(g_min, g_max,0.05)), parallel = TRUE, n.cores = 8) md <- HoltWinters(USgas, alpha = hw_grid_second$alpha, beta = hw_grid_second$beta, gamma = hw_grid_second$gamma) library(forecast) fc <- forecast(md, h = 60) plot_forecast(fc) ## End(Not run)## Not run: data(USgas) # Starting with a shallow search (sequence between 0 and 1 with jumps of 0.1) # To speed up the process, will set the parallel option to TRUE # to run the search in parallel using 8 cores hw_grid_shallow <- ts_grid(ts.obj = USgas, periods = 6, model = "HoltWinters", optim = "MAPE", window_space = 6, window_test = 12, hyper_params = list(alpha = seq(0.01, 1,0.1), beta = seq(0.01, 1,0.1), gamma = seq(0.01, 1,0.1)), parallel = TRUE, n.cores = 8) # Use the parameter range of the top 20 models # to set a narrow but more agressive search a_min <- min(hw_grid_shallow$grid_df$alpha[1:20]) a_max <- max(hw_grid_shallow$grid_df$alpha[1:20]) b_min <- min(hw_grid_shallow$grid_df$beta[1:20]) b_max <- max(hw_grid_shallow$grid_df$beta[1:20]) g_min <- min(hw_grid_shallow$grid_df$gamma[1:20]) g_max <- max(hw_grid_shallow$grid_df$gamma[1:20]) hw_grid_second <- ts_grid(ts.obj = USgas, periods = 6, model = "HoltWinters", optim = "MAPE", window_space = 6, window_test = 12, hyper_params = list(alpha = seq(a_min, a_max,0.05), beta = seq(b_min, b_max,0.05), gamma = seq(g_min, g_max,0.05)), parallel = TRUE, n.cores = 8) md <- HoltWinters(USgas, alpha = hw_grid_second$alpha, beta = hw_grid_second$beta, gamma = hw_grid_second$gamma) library(forecast) fc <- forecast(md, h = 60) plot_forecast(fc) ## End(Not run)
Heatmap plot for time series object by it periodicity (currently support only daily, weekly, monthly and quarterly frequencies)
ts_heatmap(ts.obj, last = NULL, wday = TRUE, color = "Blues", title = NULL, padding = TRUE)ts_heatmap(ts.obj, last = NULL, wday = TRUE, color = "Blues", title = NULL, padding = TRUE)
ts.obj |
A univariate time series object of a class "ts", "zoo", "xts", and the data frame family (data.frame, data.table, tbl, tibble, etc.) with a Date column and at least one numeric column. This function support time series objects with a daily, weekly, monthly and quarterly frequencies |
last |
An integer (optional), set a subset using only the last observations in the series |
wday |
An boolean, provides a weekday veiw for daily data (relevent only for objects with dates such as xts, zoo, data.frame, etc.) |
color |
A character, setting the color palette of the heatmap.
Corresponding to any of the RColorBrewer palette or any other arguments of the |
title |
A character (optional), set the plot title |
padding |
A boolean, if TRUE will add to the heatmap spaces between the observations |
data(USgas) ts_heatmap(USgas) # Show only the last 4 years ts_heatmap(USgas, last = 4 *12)data(USgas) ts_heatmap(USgas) # Show only the last 4 years ts_heatmap(USgas, last = 4 *12)
Returning the time series object main characteristics
ts_info(ts.obj)ts_info(ts.obj)
ts.obj |
A time series object of a class "ts", "mts", "xts", or "zoo" |
Text
# ts object data("USgas") ts_info(USgas) # mts object data("Coffee_Prices") ts_info(Coffee_Prices) # xts object data("Michigan_CS") ts_info(Michigan_CS)# ts object data("USgas") ts_info(USgas) # mts object data("Coffee_Prices") ts_info(Coffee_Prices) # xts object data("Michigan_CS") ts_info(Michigan_CS)
Visualization of series with its lags, can be used to identify a correlation between the series and it lags
ts_lags(ts.obj, lags = 1:12, margin = 0.02, Xshare = TRUE, Yshare = TRUE, n_plots = 3)ts_lags(ts.obj, lags = 1:12, margin = 0.02, Xshare = TRUE, Yshare = TRUE, n_plots = 3)
ts.obj |
A univariate time series object of a class "ts", "zoo" or "xts" |
lags |
An integer, set the lags range, by default will plot the first 12 lags |
margin |
Plotly parameter, either a single value or four values (all between 0 and 1). If four values provided, the first will be used as the left margin, the second will be used as the right margin, the third will be used as the top margin, and the fourth will be used as the bottom margin. If a single value provided, it will be used as all four margins. |
Xshare |
Plotly parameter, should the x-axis be shared amongst the subplots? |
Yshare |
Plotly parameter, should the y-axis be shared amongst the subplots? |
n_plots |
An integer, define the number of plots per row |
data(USgas) # Plot the first 12 lags (default) ts_lags(USgas) # Plot the seasonal lags for the first 4 years (hence, lag 12, 24, 36, 48) ts_lags(USgas, lags = c(12, 24, 36, 48)) # Setting the margin between the plot ts_lags(USgas, lags = c(12, 24, 36, 48), margin = 0.01)data(USgas) # Plot the first 12 lags (default) ts_lags(USgas) # Plot the seasonal lags for the first 4 years (hence, lag 12, 24, 36, 48) ts_lags(USgas, lags = c(12, 24, 36, 48)) # Setting the margin between the plot ts_lags(USgas, lags = c(12, 24, 36, 48), margin = 0.01)
Calculate the moving average (and double moving average) for time series data
ts_ma(ts.obj, n = c(3, 6, 9), n_left = NULL, n_right = NULL, double = NULL, plot = TRUE, show_legend = TRUE, multiple = FALSE, separate = TRUE, margin = 0.03, title = NULL, Xtitle = NULL, Ytitle = NULL)ts_ma(ts.obj, n = c(3, 6, 9), n_left = NULL, n_right = NULL, double = NULL, plot = TRUE, show_legend = TRUE, multiple = FALSE, separate = TRUE, margin = 0.03, title = NULL, Xtitle = NULL, Ytitle = NULL)
ts.obj |
a univariate time series object of a class "ts", "zoo" or "xts" (support only series with either monthly or quarterly frequency) |
n |
A single or multiple integers (by default using 3, 6, and 9 as inputs), define a two-sides moving averages by setting the number of past and future to use in each moving average window along with current observation. |
n_left |
A single integer (optional argument, default set to NULL), can be used, along with the n_right argument, an unbalanced moving average. The n_left defines the number of lags to includes in the moving average. |
n_right |
A single integer (optional argument, default set to NULL), can be used, along with the n_left argument, to set an unbalanced moving average. The n_right defines the number of negative lags to includes in the moving average. |
double |
A single integer, an optional argument. If not NULL (by default), will apply a second moving average process on the initial moving average output |
plot |
A boolean, if TRUE will plot the results |
show_legend |
A boolean, if TRUE will show the plot legend |
multiple |
A boolean, if TRUE (and n > 1) will create multiple plots, one for each moving average degree. By default is set to FALSE |
separate |
A boolean, if TRUE will separate the orignal series from the moving average output |
margin |
A numeric, set the plot margin when using the multiple or/and separate option, default value is 0.03 |
title |
A character, if not NULL (by default), will use the input as the plot title |
Xtitle |
A character, if not NULL (by default), will use the input as the plot x - axis title |
Ytitle |
A character, if not NULL (by default), will use the input as the plot y - axis title |
A one-side moving averages (also known as simple moving averages) calculation for Y[t] (observation Y of the series at time t):
MA[t|n] = (Y[t-n] + Y[t-(n-1)] +...+ Y[t]) / (n + 1),
where n defines the number of consecutive observations to be used on each rolling window along with the current observation
Similarly, a two-sided moving averages with an order of (2*n + 1) for Y[t]:
MA[t|n] = (Y[t-n] + Y[t-(n-1)] +...+ Y[t] +...+ Y[t+(n-1)] + Y[t+n]) / (2*n + 1)
Unbalanced moving averages with an order of (k1 + k2 + 1) for observation Y[t]:
MA[t|k1 & k2] = (Y[t-k1] + Y[t-(k1-1)] +...+ Y[t] +...+ Y[t+(k2-1)] + Y[t+k2]) / (k1 + k2 + 1)
The unbalanced moving averages is a special case of two-sides moving averages, where k1 and k2 represent the number of past and future periods, respectively to be used in each rolling window, and k1 != k2 (otherwise it is a normal two-sided moving averages function)
A list with the original series, the moving averages outputs and the plot
## Not run: # A one-side moving average order of 7 USgas_MA7 <- ts_ma(USgas, n_left = 6, n = NULL) # A two-sided moving average order of 13 USgas_two_side_MA <- ts_ma(USgas, n = 6) # Unbalanced moving average of order 12 USVSales_MA12 <- ts_ma(USVSales, n_left = 6, n_right = 5, n = NULL, title = "US Monthly Total Vehicle Sales - MA", Ytitle = "Thousand of Units") # Adding double MA of order 2 to balanced the series: USVSales_MA12 <- ts_ma(USVSales, n_left = 6, n_right = 5, n = NULL, double = 2, title = "US Monthly Total Vehicle Sales - MA", Ytitle = "Thousand of Units") # Adding several types of two-sided moving averages along with the unblanced # Plot each on a separate plot USVSales_MA12 <- ts_ma(USVSales, n_left = 6, n_right = 5, n = c(3, 6, 9), double = 2, multiple = TRUE, title = "US Monthly Total Vehicle Sales - MA", Ytitle = "Thousand of Units") ## End(Not run)## Not run: # A one-side moving average order of 7 USgas_MA7 <- ts_ma(USgas, n_left = 6, n = NULL) # A two-sided moving average order of 13 USgas_two_side_MA <- ts_ma(USgas, n = 6) # Unbalanced moving average of order 12 USVSales_MA12 <- ts_ma(USVSales, n_left = 6, n_right = 5, n = NULL, title = "US Monthly Total Vehicle Sales - MA", Ytitle = "Thousand of Units") # Adding double MA of order 2 to balanced the series: USVSales_MA12 <- ts_ma(USVSales, n_left = 6, n_right = 5, n = NULL, double = 2, title = "US Monthly Total Vehicle Sales - MA", Ytitle = "Thousand of Units") # Adding several types of two-sided moving averages along with the unblanced # Plot each on a separate plot USVSales_MA12 <- ts_ma(USVSales, n_left = 6, n_right = 5, n = c(3, 6, 9), double = 2, multiple = TRUE, title = "US Monthly Total Vehicle Sales - MA", Ytitle = "Thousand of Units") ## End(Not run)
Visualization functions for time series object
ts_plot(ts.obj, line.mode = "lines", width = 2, dash = NULL, color = NULL, slider = FALSE, type = "single", Xtitle = NULL, Ytitle = NULL, title = NULL, Xgrid = FALSE, Ygrid = FALSE)ts_plot(ts.obj, line.mode = "lines", width = 2, dash = NULL, color = NULL, slider = FALSE, type = "single", Xtitle = NULL, Ytitle = NULL, title = NULL, Xgrid = FALSE, Ygrid = FALSE)
ts.obj |
A univariate or multivariate time series object of class "ts", "mts", "zoo", "xts", or any data frame object with a minimum of one numeric column and either a Date or POSIXt class column |
line.mode |
A plotly argument, define the plot type, c("lines", "lines+markers", "markers") |
width |
An Integer, define the plot width, default is set to 2 |
dash |
A plotly argument, define the line style, c(NULL, "dot", "dash") |
color |
The color of the plot, support both name and expression |
slider |
Logic, add slider to modify the time axis (default set to FALSE) |
type |
A character, optional, if having multiple tims series object, will plot all series in one plot when set to "single" (default), or plot each series on a separate plot when set to "multiple" |
Xtitle |
A character, set the X axis title, default set to NULL |
Ytitle |
A character, set the Y axis title, default set to NULL |
title |
A character, set the plot title, default set to NULL |
Xgrid |
Logic,show the X axis grid if set to TRUE |
Ygrid |
Logic,show the Y axis grid if set to TRUE |
data(USVSales) ts_plot(USVSales) # adding slider ts_plot(USVSales, slider = TRUE)data(USVSales) ts_plot(USVSales) # adding slider ts_plot(USVSales, slider = TRUE)
Polor plot for time series object (ts, zoo, xts), currently support only monthly and quarterly frequency
ts_polar(ts.obj, title = NULL, width = 600, height = 600, left = 25, right = 25, top = 25, bottom = 25)ts_polar(ts.obj, title = NULL, width = 600, height = 600, left = 25, right = 25, top = 25, bottom = 25)
ts.obj |
A univariate time series object of a class "ts", "zoo" or "xts" (support only series with either monthly or quarterly frequency) |
title |
Add a title for the plot, default set to NULL |
width |
The widht of the plot in pixels, default set to 600 |
height |
The height of the plot pixels, default set to 600 |
left |
Set the left margin of the plot in pixels, default set to 25 |
right |
Set the right margin of the plot in pixels, default set to 25 |
top |
Set the top margin of the plot in pixels, default set to 25 |
bottom |
Set the bottom margin of the plot in pixels, default set to 25 |
data(USgas) ts_polar(USgas)data(USgas) ts_polar(USgas)
A quantile plot of time series data, allows the user to display a quantile plot of a series by a subset period
ts_quantile(ts.obj, upper = 0.75, lower = 0.25, period = NULL, n = 1, title = NULL, Xtitle = NULL, Ytitle = NULL)ts_quantile(ts.obj, upper = 0.75, lower = 0.25, period = NULL, n = 1, title = NULL, Xtitle = NULL, Ytitle = NULL)
ts.obj |
A univariate time series object of a class "zoo", "xts", or data frame family ("data.frame", "data.table", "tbl") |
upper |
A numeric value between 0 and 1 (excluding 0, and greater than the "lower" argument) set the upper bound of the quantile plot
(using the "probs" argument of the |
lower |
A numeric value between 0 and 1 (excluding 1, and lower than the "upper" argument) set the upper bound of the quantile plot
(using the "probs" argument of the |
period |
A character, set the period level of the data for the quantile calculation and plot representation. Must be one level above the input frequency (e.g., an hourly data can represent by daily, weekdays, monthly, quarterly and yearly). Possible options c("daily", "weekdays", "monthly", "quarterly", "yearly") |
n |
An integer, set the number of plots rows to display (by setting the nrows argument in the |
title |
A character, set the plot title, default set to NULL |
Xtitle |
A character, set the X axis title, default set to NULL |
Ytitle |
A character, set the Y axis title, default set to NULL |
## Not run: # Loading the UKgrid package to pull a multie seasonality data require(UKgrid) UKgrid_half_hour <- extract_grid(type = "xts", aggregate = NULL) # Plotting the quantile of the UKgrid dataset # No period subset ts_quantile(UKgrid_half_hour, period = NULL, title = "The UK National Grid Net Demand for Electricity - Quantile Plot") # Plotting the quantile of the UKgrid dataset # Using a weekday subset ts_quantile(UKgrid_half_hour, period = "weekdays", title = "The UK National Grid Net Demand for Electricity - by Weekdays") # Spacing the plots by setting the # number of rows of the plot to 2 ts_quantile(UKgrid_half_hour, period = "weekdays", title = "The UK National Grid Net Demand for Electricity - by Weekdays", n = 2) ## End(Not run)## Not run: # Loading the UKgrid package to pull a multie seasonality data require(UKgrid) UKgrid_half_hour <- extract_grid(type = "xts", aggregate = NULL) # Plotting the quantile of the UKgrid dataset # No period subset ts_quantile(UKgrid_half_hour, period = NULL, title = "The UK National Grid Net Demand for Electricity - Quantile Plot") # Plotting the quantile of the UKgrid dataset # Using a weekday subset ts_quantile(UKgrid_half_hour, period = "weekdays", title = "The UK National Grid Net Demand for Electricity - by Weekdays") # Spacing the plots by setting the # number of rows of the plot to 2 ts_quantile(UKgrid_half_hour, period = "weekdays", title = "The UK National Grid Net Demand for Electricity - by Weekdays", n = 2) ## End(Not run)
Transform time series object into data frame format
ts_reshape(ts.obj, type = "wide", frequency = NULL)ts_reshape(ts.obj, type = "wide", frequency = NULL)
ts.obj |
a univariate time series object of a class "ts", "zoo", "xts", and the data frame family (data.frame, data.table, tbl, tibble, etc.) with a Date column and at least one numeric column. This function support time series objects with a daily, weekly, monthly or quarterly frequencies |
type |
The reshape type - "wide" set the years as the columns and the cycle units (months or quarter) as the rows, or "long" split the time object to year, cycle unit and value |
frequency |
An integer, define the series frequency when more than one option is avaiable and the input is one of the data frame family. If set to NULL will use the first option by default when applicable - daily = c(7, 365) |
data(USgas) USgas_df <- ts_reshape(USgas)data(USgas) USgas_df <- ts_reshape(USgas)
Visualize time series object by it periodicity, currently support time series with daily, monthly and quarterly frequency
ts_seasonal(ts.obj, type = "normal", title = NULL, Ygrid = TRUE, Xgrid = TRUE, last = NULL, palette = "Set1", palette_normal = "viridis")ts_seasonal(ts.obj, type = "normal", title = NULL, Ygrid = TRUE, Xgrid = TRUE, last = NULL, palette = "Set1", palette_normal = "viridis")
ts.obj |
Input object, either a univariate time series object of a class "ts", "zoo", "xts", or a data frame object of a class "data.frame", "tbl", "data.table" as long as there is at least one "Date"/"POSIXt" and a "numeric" objects (if there are more then one, by defualt will use the first of each). Currently support only daily, weekly, monthly, and quarterly frequencies |
type |
The type of the seasonal plot - "normal" to split the series by full cycle units, or "cycle" to split by cycle units (applicable only for monthly and quarterly data), or "box" for box-plot by cycle units, or "all" for all the three plots together |
title |
Plot title - Character object |
Ygrid |
Logic,show the Y axis grid if set to TRUE (default) |
Xgrid |
Logic,show the X axis grid if set to TRUE (defualt) |
last |
Subset the data to the last number of observations |
palette |
A character, the color palette to be used when the "cycle" or "box" plot are being selected (by setting the type to "cycle", "box", or "all"). All the palettes in the RColorBrewer and viridis packages are available to be use, the default option is "Set1" from the RColorBrewer package |
palette_normal |
A character, the color palette to be used when the "normal" plot is being selected (by setting the type to "normal" or "all"). All the palettes in the RColorBrewer and viridis packages are available to be used, the default palette is "viridis" from the RColorBrewer package |
data(USgas) ts_seasonal(USgas) # Seasonal box plot ts_seasonal(USgas, type = "box") # Plot all the types ts_seasonal(USgas, type = "all")data(USgas) ts_seasonal(USgas) # Seasonal box plot ts_seasonal(USgas, type = "box") # Plot all the types ts_seasonal(USgas, type = "all")
Split a time series object into training and testing partitions
ts_split(ts.obj, sample.out = NULL)ts_split(ts.obj, sample.out = NULL)
ts.obj |
A univariate time series object of a class "ts" or "tsibble" |
sample.out |
An integer, set the number of periods of the testing or sample out partition, defualt set for 30 percent of the lenght of the series |
## Split the USgas dataset into training and testing partitions ## Set the last 12 months as a testing partition ## and the rest as a training partition data(USgas, package = "TSstudio") split_USgas <- ts_split(ts.obj = USgas, sample.out = 12) training <- split_USgas$train testing <- split_USgas$test length(USgas) length(training) length(testing)## Split the USgas dataset into training and testing partitions ## Set the last 12 months as a testing partition ## and the rest as a training partition data(USgas, package = "TSstudio") split_USgas <- ts_split(ts.obj = USgas, sample.out = 12) training <- split_USgas$train testing <- split_USgas$test length(USgas) length(training) length(testing)
A row sum function for multiple time series object ("mts"), return the the summation of the "mts" object as a "ts" object
ts_sum(mts.obj)ts_sum(mts.obj)
mts.obj |
A multivariate time series object of a class "mts" |
x <- matrix(c(1:100, 1:100, 1:100), ncol = 3) mts.obj <- ts(x, start = c(2000, 1), frequency = 12) ts_total <- ts_sum(mts.obj)x <- matrix(c(1:100, 1:100, 1:100), ncol = 3) mts.obj <- ts(x, start = c(2000, 1), frequency = 12) ts_total <- ts_sum(mts.obj)
3D surface plot for time series object by it periodicity (currently support only monthly and quarterly frequency)
ts_surface(ts.obj)ts_surface(ts.obj)
ts.obj |
a univariate time series object of a class "ts", "zoo" or "xts" (support only series with either monthly or quarterly frequency) |
ts_surface(USgas)ts_surface(USgas)
Transform a time series object to Prophet data frame input format
ts_to_prophet(ts.obj, start = NULL)ts_to_prophet(ts.obj, start = NULL)
ts.obj |
A univariate time series object of a class "ts", "zoo", "xts", with a daily, weekly, monthly , quarterly or yearly frequency |
start |
A date object (optional), if the starting date of the series is known. Otherwise, the date would be derive from the series index |
A data frame object
data(USgas) ts_to_prophet(ts.obj = USgas) # If known setting the start date of the input object ts_to_prophet(ts.obj = USgas, start = as.Date("2000-01-01"))data(USgas) ts_to_prophet(ts.obj = USgas) # If known setting the start date of the input object ts_to_prophet(ts.obj = USgas, start = as.Date("2000-01-01"))
Monthly total vehicle sales and unemployment rate: 1976 - 2019. Units: Dollars per Kg
US_indicatorsUS_indicators
Time series data - 'data.frame' object
U.S. Bureau of Economic Analysis, Total Vehicle Sales [TOTALNSA], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/TOTALNSA, January 7, 2018. U.S. Bureau of Labor Statistics, Civilian Unemployment Rate [UNRATENSA], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/UNRATENSA, January 6, 2018.
ts_plot(US_indicators)ts_plot(US_indicators)
US monthly natural gas consumption: 2000 - 2019. Units: Billion Cubic Feet
USgasUSgas
Time series data - 'ts' object
U.S. Bureau of Transportation Statistics, Natural Gas Consumption [NATURALGAS], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/NATURALGAS, January 7, 2018.
ts_plot(USgas) ts_seasonal(USgas, type = "all")ts_plot(USgas) ts_seasonal(USgas, type = "all")
US monthly civilian unemployment rate: 1948 - 2019. Units: Percent
USUnRateUSUnRate
Time series data - 'ts' object
U.S. Bureau of Labor Statistics, Civilian Unemployment Rate [UNRATENSA], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/UNRATENSA, January 6, 2018.
ts_plot(USUnRate) ts_seasonal(USUnRate)ts_plot(USUnRate) ts_seasonal(USUnRate)
US monthly total vehicle sales: 1976 - 2019. Units: Thousands of units
USVSalesUSVSales
Time series data - 'ts' object
U.S. Bureau of Economic Analysis, Total Vehicle Sales [TOTALNSA], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/TOTALNSA, January 7, 2018.
ts_plot(USVSales) ts_seasonal(USVSales)ts_plot(USVSales) ts_seasonal(USVSales)
Converting 'xts' object to 'ts' object
xts_to_ts(xts.obj, frequency = NULL, start = NULL)xts_to_ts(xts.obj, frequency = NULL, start = NULL)
xts.obj |
A univariate 'xts' object |
frequency |
A character, optional, if not NULL (default) set the frequency of the series |
start |
A Date or POSIXct/lt object, optional, can be used to set the starting date or time of the series |
data(Michigan_CS) class(Michigan_CS) ts_plot(Michigan_CS) Michigan_CS_ts <- xts_to_ts(Michigan_CS) ts_plot(Michigan_CS_ts) # Defining the frequency and starting date of the series Michigan_CS_ts1 <- xts_to_ts(Michigan_CS, start = as.Date("1980-01-01"), frequency = 12 ) ts_plot(Michigan_CS_ts1)data(Michigan_CS) class(Michigan_CS) ts_plot(Michigan_CS) Michigan_CS_ts <- xts_to_ts(Michigan_CS) ts_plot(Michigan_CS_ts) # Defining the frequency and starting date of the series Michigan_CS_ts1 <- xts_to_ts(Michigan_CS, start = as.Date("1980-01-01"), frequency = 12 ) ts_plot(Michigan_CS_ts1)
Converting 'zoo' object to 'ts' object
zoo_to_ts(zoo.obj)zoo_to_ts(zoo.obj)
zoo.obj |
a univariate 'zoo' object |
data("EURO_Brent", package = "TSstudio") class(EURO_Brent) ts_plot(EURO_Brent) EURO_Brent_ts <- zoo_to_ts(EURO_Brent) class(EURO_Brent_ts) ts_plot(EURO_Brent_ts)data("EURO_Brent", package = "TSstudio") class(EURO_Brent) ts_plot(EURO_Brent) EURO_Brent_ts <- zoo_to_ts(EURO_Brent) class(EURO_Brent_ts) ts_plot(EURO_Brent_ts)