13-final-thoughts.qmd

---
title: "ETC3550/ETC5550 Applied forecasting"
author: "Some final thoughts"
institute: "OTexts.org/fpp3/"
pdf-engine: pdflatex
fig-width: 7.5
fig-height: 3.5
format:
  beamer:
    theme: monash
    aspectratio: 169
    fontsize: 14pt
    section-titles: false
    knitr:
      opts_chunk:
        dev: "cairo_pdf"
include-in-header: header.tex
execute:
  echo: false
  message: false
  warning: false
---

```{r setup, include=FALSE}
source("setup.R")
```

# Assignment 1

## Assignment 1

**Stock price forecasting** (Q1 and Q5)

* Hard to beat naive forecast
* Random walk model says forecast variance = $h\sigma^2$.

\pause

**Maximum temperature at Melbourne airport on 12 April 2021.** (Q2)

* Weather is relatively stationary over similar time of year and recent years.
* So take mean and var of max temp in April over last 10 years.

## Assignment 1

**Difference in points in AFL match** (Q3)

* Teams vary in strength from year to year.
* Could look at distribution of for-against points from 2020 across all games for each team. Assume distributions independent.

\pause

**Seasonally adjusted estimate of total employment** (Q4)

* Probably locally trended.
* Perhaps use drift method based on average monthly change in last 2 years.

# Some case studies

## CASE STUDY 1: Paperware company

\fontsize{11.5}{13}\sf

\begin{textblock}{7.6}(0.2,1.5)
\textbf{Problem:} Want forecasts of each of hundreds of
items. Series can be stationary, trended or seasonal. They currently
have a large forecasting program written in-house but it doesn't seem
to produce sensible forecasts. They want me to tell them what is
wrong and fix it.

\vspace*{0.cm}

\textbf{Additional information}\vspace*{-0.2cm}\fontsize{11.5}{13}\sf
\begin{itemize}\itemsep=0cm\parskip=0cm
\item  Program  written in COBOL making numerical calculations limited. It is not possible to do any optimisation.
\item Their programmer has little experience in numerical computing.
\item They employ no statisticians and want the program to produce forecasts \rlap{automatically.}
\end{itemize}
\end{textblock}

\placefig{8}{1.4}{width=4.8cm}{tableware2}

## CASE STUDY 1: Paperware company

### Methods currently used
A
: 12 month average

C
: 6 month average

E
: straight line regression over last 12 months

G
: straight line regression over last 6 months

H
: average slope between last year's and this year's values.
  (Equivalent to differencing at lag 12 and taking mean.)

I
: Same as H except over 6 months.

K
: I couldn't understand the explanation.

## CASE STUDY 2: PBS

\fontsize{13.5}{15.5}\sf

  * In 2001: \$4.5 billion budget, under-forecasted by \$800 million.
  * Thousands of products. Seasonal demand.
  * Subject to covert marketing, volatile products, uncontrollable expenditure.
  * Although monthly data available for 10 years, data are aggregated to annual values, and only the first three years are used in estimating the forecasts.
  * All forecasts being done with the \texttt{FORECAST} function in MS-Excel!

\begin{alertblock}<2>{}
\textbf{Problem:} How to do the forecasting better?
\end{alertblock}
\vspace{5cm}

## CASE STUDY 3: Car fleet company

**Client:** One of Australia's largest car fleet companies

**Problem:** how to forecast resale value of vehicles? How
should this affect leasing and sales policies?

\pause

### Additional information
 - They can provide a large amount of data on previous vehicles and their eventual resale values.
 - The resale values are currently estimated by a group of specialists. They see me as a threat and do not cooperate.

## CASE STUDY 4: Airline

**Problem:** how to forecast passenger traffic on major routes?

```{r, fig.height=3.5, echo=FALSE}
ansett |>
  filter(Airports == "MEL-SYD") |>
  autoplot(Passengers) +
  labs(
    title = "Economy class passengers: Melbourne-Sydney",
    y = "Thousands"
  )
```

\vspace*{10cm}

## CASE STUDY 4: Airline

**Problem:** how to forecast passenger traffic on major routes?

### Additional information

  * They can provide a large amount of data on previous routes.
  * Traffic is affected by school holidays, special events such as
the Grand Prix, advertising campaigns, competition behaviour, etc.
  * They have a highly capable team of people who are able to do
most of the computing.

# Exam

## Exam: 9.00am (AEST) 7 June

Five questions, all to be attempted.

A
: Short answers/explanations. Write about 1/4 page on four topics (out of six possible topics). Nuanced answers required.

B
: Describing a time series, choosing a forecasting method

C
: ETS models

D
: ARIMA models

E
: (Dynamic) regression models

\vspace*{0.3cm}

###
  $\bullet$ Moodle Quiz\qquad $\bullet$ Open book\qquad $\bullet$ 2 hr 40 min

## Exam and R

 * Parts \alert{B}, \alert{C} and \alert{E} require interpretation of R output, but no coding.
 * Part \alert{D} requires some coding (part of the code will be provided) and interpretation of R output.
 * All R coding will be very similar to examples you have done before.
 * Submitted answers will be automatically checked for close matches.
 * Enter answers on Moodle as you go, to avoid internet issues at the end.

## Preparing for the exam

  - Exams from 2018--2020 on Moodle. Solutions to follow by Monday.
  - Exercises.  Make sure you have done them all!
  - Identify your weak points and practice them.
  - Write your own summary of the material.
  - Practice explaining the material to a class-mate.

\pause

### Help available
 * Ask on Moodle forum
 * See a tutor during the consultation times.

## Useful resources for forecasters
\fontsize{13}{12}\sf

**Organization:**

 * International Institute of Forecasters.

**Annual Conference:**

 * International Symposium on Forecasting

**Journals:**

 * International Journal of Forecasting
 * Foresight

\begin{block}{}
Links to all of the above at \textbf{forecasters.org}
\end{block}\pause
\begin{block}{}
Good forecasters are not smarter than everyone else, they merely have their ignorance better organised.
\end{block}\pause
\begin{alertblock}{}\centering
Please fill in your SETU
\end{alertblock}