Assessment Summary: Forecasting and Quantitative Analysis Overview of Assessment Requirements This assessment focuses on developing skills in data analysis, forecasting, and model evaluation u

Assignment Overview

Part A 

The aim in this part of the assignment is to understand the data, perform transformation (if required), and use simple forecasting models to produce forecasts.

Question 1

Produce appropriate plots in order to become familiar with your data. Make sure you label your axes and plots appropriately. Comment on the plots. What do you see? (50 words per plot).

Question 2

Would transforming your data be useful? If transformation is required, compare two approaches graphically. Choose the best transformation, justifying your choice (100 words).

Question 3

Apply the two most appropriate benchmark (simple forecasting) methods, justifying your choices 
Question 4

Perform a thorough residual analysis for each model. Do the residuals appear to be white noise? 

Question 5

Generate and plot forecasts and forecast intervals for the next 2 years from the two benchmark methods, also plotting the observed data. You may choose to provide a separate plot for each model (along with the observed data) for better visualisation. You may also choose to plot on a shorter period of say 5 last years for clearer visualisation. Compare and discuss your findings, commenting on the merits/limitations of either or both modelling approaches (100 words).

Part B (25 marks)

The aim in this part of the assignment is to build an ARIMA model and use it to forecast.

Question 6

Visually inspect your transformed data and decide what differencing is required to achieve stationarity. Analyse using relevant plots at every step, commenting on each plot and justifying your actions. (50 words per plot).

Question 7

Estimate an ARIMA model using the auto-ARIMA function in R. Tabulate your results.

Question 8

Perform a thorough residual diagnostics analysis for your estimated model. Discuss your results. 

Question 9

Generate and plot forecasts and forecast intervals for the next two years. Comment on the results 

Part C (20 marks)

You have now built three models with your dataset. Next, the aim is to evaluate the three models.

Question 10

Create a training set with your data by leaving two years’ worth of observations as the test set. Generate forecasts for the last two years (the period of the test set), from the three models you have estimated in Parts A (two benchmark models) & B (ARIMA model). Plot the forecasts (both point forecasts and prediction intervals) together with the observed data and comment on these (100 words). You may choose to plot on a shorter period of say 5 last years for clearer visualisation. Make sure the visualisations are clear.

Question 11

Compute the accuracy of your forecasts generated from the three models in a table. Which model does best and why? 

Assessment Summary: Forecasting and Quantitative Analysis

Overview of Assessment Requirements

This assessment focuses on developing skills in data analysis, forecasting, and model evaluation using quantitative techniques. It is divided into three key parts (A, B, and C), each designed to progressively enhance the student’s understanding of time series forecasting, model building, and validation.

• Part A emphasizes understanding the dataset, performing necessary data transformations, and applying simple benchmark forecasting models to generate forecasts.
• Part B focuses on building an ARIMA model, conducting residual diagnostics, and producing forecasts using the model.
• Part C involves evaluating and comparing the three models (two from Part A and one from Part B) using test data and accuracy measures to determine the best-performing model.

The key components covered in the assessment include:

1. Visual exploration of data using appropriate plots.
2. Data transformation and justification for the chosen method.
3. Application of benchmark forecasting models.
4. Residual and diagnostic analysis.
5. Generation and comparison of forecasts and confidence intervals.
6. Model selection and evaluation using statistical accuracy measures.