Statistics Final Year Topic: Rolling-Origin Evaluation of ARIMA Forecasts with Changing Training Windows

This Statistics final year project uses simulated series with a documented break and a held-out forecasting period to investigate a specific question in time-series forecasting. The analysis is designed around known generating conditions so that the behaviour of competing statistical procedures can be checked.

Generate the Complete Project Read project brief

Why choose this project topic?

This study makes rolling forecasts an explicit, reproducible comparison. Working with simulated series with a documented break and a held-out forecasting period lets you separate the target quantity from its estimate and explain when a statistical conclusion depends on assumptions.

How does training-window length affect short-horizon ARIMA forecast error when a time series contains a structural change?

Agree the scenario ranges, sample sizes and reporting measures for rolling forecasts before running the study.

Proposed project objectives

01Specify the data-generating conditions for simulated series with a documented break and a held-out forecasting period.
02Implement a reproducible analysis of rolling forecasts with documented software settings.
03Evaluate the estimates and uncertainty relevant to this question: How does training-window length affect short-horizon ARIMA forecast error when a time series contains a structural change?

A suggested research approach

Compare fixed and expanding windows using the same rolling origins. Select model settings using training observations only and report errors separately before and after the structural change. Write the analysis before inspecting favourable runs, record random seeds where simulation is used, and keep generated study data distinct from observed field data.

What you will need

A written design for simulated series with a documented break and a held-out forecasting period
Statistical software supporting time-series forecasting and reproducible scripts
A supervisor-agreed range of scenarios and computational budget

Keep your project scope clear

A good average forecast score can hide failure around breaks, and tuning on the final test period leaks future information.

Statistics project chapter outline

Use this outline as a starting point. You can edit the chapter titles to match your department’s format during setup.

Chapter 1Introduction
Chapter 2Literature Review
Chapter 3Theory and Methodology
Chapter 4Results and Applications
Chapter 5Summary, Conclusion and Recommendations

Turn this topic into your own final year project.

Your title, department, research question and outline are ready. Add your institution, personalise the details and continue to your project workspace.

Generate the Complete Project Generation uses your word balance. Review the draft and supply your own verified research findings.