Automatically plots with autoplot function from forecasting object

问题

I am foresting with combination of data sets from fpp2 package and forecasting function from the forecast package. Output from this forecasting is object list with SNAIVE_MODELS_ALL. This object contain data separate for two series, where first is Electricity and second is Cement. You can see code below :

#  CODE
library(fpp2)
library(dplyr)
library(forecast)
library(gridExtra)
library(ggplot2)

#INPUT DATA
mydata_qauselec <- qauselec
mydata_qcement <- window(qcement, start = 1956, end = c(2010, 2))

# Мerging data
mydata <- cbind(mydata_qauselec, mydata_qcement)
colnames(mydata) <- c("Electricity", "Cement")

# Test Extract Name
mydata1 <- data.frame(mydata)
COL_NAMES <- names(mydata1)
rm(mydata_qauselec, mydata_qcement)

# FORCASTING HORIZON
forecast_horizon <- 12

#FORCASTING
BuildForecast <- function(Z, hrz = forecast_horizon) {
  timeseries <- msts(Z, start = 1956, seasonal.periods = 4)
  forecast <- snaive(timeseries, biasadj = TRUE, h =  hrz)
}  

frc_list <- lapply(X = mydata1, BuildForecast)

#FINAL FORCASTING
SNAIVE_MODELS_ALL<-lapply(frc_list,  forecast)

So my intention here is to put this object SNAIVE_MODELS_ALL into autoplot function in order to get two plots like pic below.

With code below I draw both plots separate, but my main intention is to do this with function autoplot and some function like apply or something similar, which can automatically draw this two chart like pic above.This is only small example in real example I will have maybe 5 or 10 charts.

#PLOT 1
P_PLOT1<-autoplot(SNAIVE_Electricity,main = "Snaive Electricity forecast",xlab = "Year", ylab = "in billion kWh")+
  autolayer(SNAIVE_Electricity,series="Data")+
  autolayer(SNAIVE_Electricity$fitted,series="Forecasts")

# PLOT 2
P_PLOT2<-autoplot(SNAIVE_Cement,main = "Snaive Cement forecast",xlab = "Year", ylab = "in millions of tonnes")+
  autolayer(SNAIVE_Cement,series="Data")+
  autolayer(SNAIVE_Cement$fitted,series="Forecasts")

#UNION PLOTS (PLOT 1 AND PLOT 2)
SNAIVE_PLOT_ALL<-grid.arrange(P_PLOT1,P_PLOT2)

So can anybody help me with this code ?

回答1:

If I understand in a proper way, one of the difficulties with that problem is that each plot should have a specific title and y label. One of the possible solutions is to set the plot titles and y-lables as function arguments:

PlotForecast <- function(df_pl, main_pl, ylab_plt){
    autoplot(df_pl,
    main = main_pl,
    xlab = "Year", ylab = ylab_plt)+
    autolayer(df_pl,series="Data")+
    autolayer(df_pl$fitted,series="Forecasts")
}

Prepare lists of the plot labels to be used with PlotForecast():

main_lst <- list("Snaive Electricity forecast", "Snaive Cement forecast")
ylab_lst <- list("in billion kWh", "in millions of tonnes")

Construct a list of plot-objects using a base Map() function:

PL_list <- Map(PlotForecast, df_pl = SNAIVE_MODELS_ALL, main_pl = main_lst,
    ylab_plt= ylab_lst)

Then all we have to do is to call grid.arrange() with the plot list:

do.call(grid.arrange, PL_list)

Note, please, that main_lst and ylab_lst are created manually for demonstration purposes, but it is not the best way if you work with a lot of charts. Ideally, the labels should be generated automatically using the original SNAIVE_PLOT_ALL list.

来源：https://stackoverflow.com/questions/57693208/automatically-plots-with-autoplot-function-from-forecasting-object

标签

ggplot2

forecast