1. Load and extract data for navive and theta method from m3 yearly data

library(forvision)
ts <- m3_yearly_ts
fc <- m3_yearly_fc
af <- createAFTS(ts, fc)
naive_data <- dplyr::filter(af, method_id=="NAIVE2")
theta_data <- dplyr::filter(af, method_id=="THETA")
knitr::kable(head(naive_data))
series_id category value timestamp method_id forecast horizon origin_timestamp
Y1 MICRO 5379.75 1989 NAIVE2 4936.99 1 1988
Y1 MICRO 6158.68 1990 NAIVE2 4936.99 2 1988
Y1 MICRO 6876.58 1991 NAIVE2 4936.99 3 1988
Y1 MICRO 7851.91 1992 NAIVE2 4936.99 4 1988
Y1 MICRO 8407.84 1993 NAIVE2 4936.99 5 1988
Y1 MICRO 9156.01 1994 NAIVE2 4936.99 6 1988 
knitr::kable(head(theta_data))
series_id category value timestamp method_id forecast horizon origin_timestamp
Y1 MICRO 5379.75 1989 THETA 5414.60 1 1988
Y1 MICRO 6158.68 1990 THETA 5934.47 2 1988
Y1 MICRO 6876.58 1991 THETA 6331.96 3 1988
Y1 MICRO 7851.91 1992 THETA 6822.35 4 1988
Y1 MICRO 8407.84 1993 THETA 7140.76 5 1988
Y1 MICRO 9156.01 1994 THETA 7602.15 6 1988

2. Вычисление ошибки и построение графика

naive_data$error <- naive_data$value - naive_data$forecast
theta_data$error <- theta_data$value - theta_data$forecast
knitr::kable(head(naive_data))
series_id category value timestamp method_id forecast horizon origin_timestamp error
Y1 MICRO 5379.75 1989 NAIVE2 4936.99 1 1988 442.76
Y1 MICRO 6158.68 1990 NAIVE2 4936.99 2 1988 1221.69
Y1 MICRO 6876.58 1991 NAIVE2 4936.99 3 1988 1939.59
Y1 MICRO 7851.91 1992 NAIVE2 4936.99 4 1988 2914.92
Y1 MICRO 8407.84 1993 NAIVE2 4936.99 5 1988 3470.85
Y1 MICRO 9156.01 1994 NAIVE2 4936.99 6 1988 4219.02 
knitr::kable(head(theta_data))
series_id category value timestamp method_id forecast horizon origin_timestamp error
Y1 MICRO 5379.75 1989 THETA 5414.60 1 1988 -34.85
Y1 MICRO 6158.68 1990 THETA 5934.47 2 1988 224.21
Y1 MICRO 6876.58 1991 THETA 6331.96 3 1988 544.62
Y1 MICRO 7851.91 1992 THETA 6822.35 4 1988 1029.56
Y1 MICRO 8407.84 1993 THETA 7140.76 5 1988 1267.08
Y1 MICRO 9156.01 1994 THETA 7602.15 6 1988 1553.86

For naive method

library(ggplot2)
graph <- ggplot(naive_data, aes(x = value, y = abs(error))) +
    geom_point(colour = "blue", size = 1.5) +
    stat_smooth(method = "loess",
        col = "#C42126",
        size = 1.5)
graph

For theta method

library(ggplot2)
graph <- ggplot(theta_data, aes(x = value, y = abs(error))) +
    geom_point(colour = "blue", size = 1.5) +
    stat_smooth(method = "loess",
        col = "#C42126",
        size = 1.5)
graph

for naive and theta methods

data <- rbind(naive_data, theta_data)
library(ggplot2)
graph <- ggplot(data, aes(x = value, y = abs(error), shape = method_id, colour = method_id)) +
    geom_point(size = 1.5) +
    stat_smooth(method = "loess")
graph