Spirit Airlines

Import and Examine Data

# data source: https://www.transtats.bts.gov/Data_Elements.aspx?Data=1

spirit <- read_csv("data/spirit_passengers_5_2_2024.csv",
                   show_col_types = F, skip=1)|>
  filter(Month != "TOTAL") |>
  mutate(date_som = ym(paste(Year, Month)),
         Date = ceiling_date(date_som, "month")-1,
         Total = (TOTAL/1000000) |> round(2)) |>
  select(Date, Total) |>
  glimpse(width=60)

Rows: 256
Columns: 2
$ Date  <date> 2002-10-31, 2002-11-30, 2002-12-31, 2003-01…
$ Total <dbl> 0.30, 0.33, 0.36, 0.34, 0.31, 0.36, 0.35, 0.…

spirit_pp <- spirit |>          # post_pandemic data 
  filter(Date >= "2021-06-01") |>
  glimpse(width=60)

Rows: 32
Columns: 2
$ Date  <date> 2021-06-30, 2021-07-31, 2021-08-31, 2021-09…
$ Total <dbl> 2.92, 3.39, 2.43, 2.47, 2.72, 2.92, 3.00, 2.…

Interactive Time Series Plot

Full and Truncated Data Plots

Plot of Full Time Series

Seasonal pattern was disrupted by the pandemic when air travel was considered dangerous.

Plot of Truncated Time Series

Once vaccines became readily available, air travel began to resume it’s traditional pattern:

• Peaks occur at the end of July

• Low points occur at the end of Jnauary

• Pattern is not straightforward to discern because post-pandemic data are fairly limited.

• Data for February and March of 2024 are not available yet.

Forecast Modeling

Create Time Series (ts)

spirit_ts <- ts(spirit_pp$Total, freq=12, start=c(2021,6)) # create time series

spirit_ts # display time series

      Jan  Feb  Mar  Apr  May  Jun  Jul  Aug  Sep  Oct  Nov  Dec
2021                          2.92 3.39 2.43 2.47 2.72 2.92 3.00
2022 2.55 2.58 3.37 3.00 3.37 3.33 3.50 3.38 3.08 3.43 3.38 3.44
2023 3.41 3.26 3.92 3.71 3.81 3.64 3.79 3.69 3.29 3.84 3.81 3.86
2024 3.33                                                       

Model 1

spirit_model1 <- spirit_ts |> auto.arima(ic="aic", seasonal=F)

spirit_forecast1 <- spirit_model1 |> forecast(h=12)

(spirit_fcp1 <- autoplot(spirit_forecast1) + 
    labs(y = "Number of Passenger (Mill.)") + 
    theme_classic())

spirit_forecast1

         Point Forecast    Lo 80    Hi 80    Lo 95    Hi 95
Feb 2024       3.481283 3.116950 3.845616 2.924084 4.038482
Mar 2024       3.847076 3.452806 4.241346 3.244092 4.450060
Apr 2024       3.700524 3.288744 4.112304 3.070761 4.330287
May 2024       3.581390 3.149308 4.013472 2.920578 4.242203
Jun 2024       3.391083 2.908536 3.873631 2.653090 4.129076
Jul 2024       3.679531 3.101981 4.257081 2.796245 4.562817
Aug 2024       3.777295 3.186909 4.367680 2.874378 4.680211
Sep 2024       3.576580 2.965826 4.187333 2.642513 4.510647
Oct 2024       3.511578 2.868117 4.155040 2.527489 4.495667
Nov 2024       3.541708 2.849222 4.234193 2.482643 4.600772
Dec 2024       3.738095 3.008390 4.467799 2.622108 4.854081
Jan 2025       3.659739 2.919221 4.400257 2.527215 4.792263

checkresiduals(spirit_forecast1, test=F)

(acr1 <- accuracy(spirit_forecast1)) 

                   ME      RMSE      MAE       MPE     MAPE     MASE
Training set 0.037663 0.2562558 0.198798 0.6831742 6.154862 0.423425
                    ACF1
Training set 0.006136755

Model 2

spirit_model2 <- spirit_ts |> auto.arima(ic="aic", seasonal=T) 

spirit_forecast2 <- spirit_model2 |> forecast(h=12)

(spirit_fcp2 <- autoplot(spirit_forecast2) + 
    labs(y = "Number of Passenger (Mill.)") + 
    theme_classic())

spirit_forecast2

         Point Forecast    Lo 80    Hi 80    Lo 95    Hi 95
Feb 2024        3.50522 3.184490 3.825950 3.014706 3.995734
Mar 2024        4.16522 3.829138 4.501302 3.651227 4.679213
Apr 2024        3.95522 3.604457 4.305983 3.418775 4.491666
May 2024        4.05522 3.690367 4.420074 3.497225 4.613215
Jun 2024        3.88522 3.506800 4.263640 3.306477 4.463963
Jul 2024        4.03522 3.643704 4.426736 3.436448 4.633993
Aug 2024        3.93522 3.531032 4.339409 3.317067 4.553373
Sep 2024        3.53522 3.118745 3.951696 2.898276 4.172164
Oct 2024        4.08522 3.656810 4.513630 3.430023 4.740417
Nov 2024        4.05522 3.615199 4.495241 3.382266 4.728175
Dec 2024        4.10522 3.653886 4.556554 3.414965 4.795475
Jan 2025        3.57522 3.112851 4.037590 2.868087 4.282353

checkresiduals(spirit_forecast2, test=F)

(acr2 <- accuracy(spirit_forecast2)) 

                      ME      RMSE       MAE        MPE     MAPE     MASE
Training set -0.01674151 0.1877002 0.1133833 -0.4391385 3.251134 0.241498
                    ACF1
Training set -0.02364004