Individual Assignment
Forecasting Global Gold Demand Using Gold Prices and Macroeconomic Indicators Due
Overview
In this assignment, you will apply forecasting techniques to a real-world economic problem: predicting quarterly global gold demand using macroeconomic indicators such as gold prices, inflation rates, interest rates, and the US dollar index. You will collect and clean time series data, build models using regression with ARIMA errors, explore ensemble and bootstrap-based methods, and evaluate the forecast accuracy of multiple approaches.
The assignment reflects the analytical tasks faced by economists, financial analysts, policymakers, and commodity traders. Your final submission will take the form of a professional forecasting report, including model diagnostics, economic interpretation, and policy implications.
Stuck in This Assignment? Deadlines Are Near?
Learning Objectives
Upon completing this assignment, you will be able to:
- Apply time series regression techniques, including regression with ARIMA errors, to model economic demand.
- Source, clean, and prepare economic and financial time series data using appropriate transformations, interpolation, and outlier treatment.
- Diagnose and validate forecast models using statistical tools such as ACF/PACF, residual analysis, and model selection criteria.
- Develop ensemble forecasts and bootstrap-based bagged models to improve forecast robustness and accuracy.
- Evaluate model performance using forecast accuracy metrics and justify model choice based on statistical and economic reasoning.
- Communicate forecasting results effectively in a structured report format, integrating data visualisations, economic insights, and stakeholder relevance.
Important Instructions and Clarifications
Weight: 40% of your final grade
Assignment 3 assesses your understanding of all topics covered this semester, with a particular focus on Weeks 7–11. All materials required to complete this assessment—lecture videos, lecture slides, tutorial guides
Please read all instructions carefully. Failure to follow the requirements may result in loss of marks.
- Coverage of Topics
This assessment evaluates:
- Your understanding of core concepts from Weeks 7–11
- Your analytical and interpretation skills
- Your ability to apply models to the specific case provided
- Proper coding, reasoning, and communication of results
- Submission Format Requirements
You must submit a single PDF file, which includes:
- Your answers to every question
- Screenshots of your R code
- Outputs, graphs, and tables
- Clear explanations and interpretations linked to your own analysis
Do NOT submit multiple files or upload raw R scripts.
- How to Answer the Questions
- Your answers must be case-specific, based on the scenario in the assignment.
- Generic answers (e.g., definitions copied from the internet or AI tools) will not receive marks.
- You must interpret your own results (plots, ACF/PACF, regression, ARIMA output, etc.).
- Your written explanations must demonstrate your understanding of the methods.
Any answer detected as AI-generated will receive 0 marks.
- Formatting Guidelines (Required)
Please follow these formatting rules strictly:
- Font: Arial
- Font size: 11
- Line spacing:5
- Alignment:
o Justified text
- Graphs:
o Must be clearly labelled
o Axes
must be named
- Titles must be descriptive o Figures must be neat, readable, and properly formatted
Incorrect formatting may result in a deduction of marks.
- Academic Integrity
- Your submission must be uploaded through Turnitin.
- The similarity index must be below 35%.
- Plagiarism, copying, and academic misconduct will be reported to the university.
- Identical or highly similar submissions between students will result in 0 marks for all involved.
- Clarification Policy
You may email me only to clarify the meaning of a question.
You MUST NOT email me to ask:
- whether your answer is correct
- whether your output is correct
- whether your method is correct
This is your assignment. Your analytical choices and reasoning are part of the evaluation.
- Submission Instructions
- Submit your single PDF to the Moodle submission link before the due date.
- Late submissions will follow the school’s standard late penalty policy.
- Ensure all figures, answers, and codes are visible in the PDF.
- Ensure the file opens correctly and is not corrupted.
- Final Reminder
Your assignment should demonstrate:
- Your understanding of course material
- Your ability to apply statistical and time series methods
- Your own critical thinking and interpretation
- Professional presentation and formatting
Assignments that do not meet the requirements above will lose marks accordingly.
Get 30% Discount on This Assignment Answer Today!
A. Data Collection and Preprocessing (10m)
- Collect and merge data from the required sources.
- Gold Demand (https://www.gold.org/goldhub/data/gold-demand-by-country)
○ Gold Price (LBMA, WGC, FRED)
○ Inflation (CPI or PPI)
○ Interest Rates (Fed Funds Rate, ECB rate)
○ USD Index (DXY or USD/EUR)
- Clean the data using:
- interp() for missing values
○ tsclean() for outliers
○ diff() and ADF tests for stationarity
Questions:
- What transformations were needed to make your variables stationary?
- Are there seasonal patterns or structural breaks in the gold demand series?
- How did outliers affect the gold demand series? Did cleaning them change trends?
B. Regression with ARIMA Errors (40m)
Fit a model. Main regression equation:
ARIMA error structure:
Use auto.arima(…, xreg=…) in R.
Questions:
- Which predictor had the strongest coefficient? Was its sign expected (Provide interpretation)? Is the model overall significant?
- Are the residuals from your model white noise? What does this imply?
- Is an ARIMA error structure necessary for this model? Explain clearly and present the full model if an ARIMA error structure model was built.
- Is the forecasting output reasonable?
C. Advanced Forecasting Methods (30m)
1. Forecast Combination (Ensemble)
The time series of interest will be GoldDemandt from section B.
Use an appropriate simple forecasting method, auto ETS, auto ARIMA, and STL to forecast and then average them.
Questions:
- Did the ensemble model perform better than individual models? Why might that be?
- What are the benefits of combining forecasts in volatile markets like gold?
2. Bootstrap & Bagging
Use baggedModel() or baggedETS() to generate bagged forecasts from bootstrapped series. The
number of bootstrap series that you should produce is B = 20.
- How does bootstrapping help capture uncertainty in your time series?
- Did bagging reduce your forecast error? In what types of series is bagging most useful?
D. Model Evaluation and Selection (10m)
Ignoring the residual performance, forecasting interval, compare the performance of the model for GoldDemandt in Section B, Section C-a, and Section C-b using:
- RMSE, MAE, MAPE
- Cross-validation (CV), set h = 12
Questions:
- Which model had the lowest out-of-sample forecast error? E.
Interpretation and Reflection (10m)
Questions:
Section B required a model built focusing on causality, while Section C demanded a forecasting output that was purely predictive. Provide a reflection on the merits of each method and its respective disadvantages. Provide examples of data that is potentially better off building predictions based on causality and models that are purely predictive.
End of Assignment.
Get 30% Discount on This Assignment Answer Today!
Get Help By Expert
Struggling to complete your ETW3420: Principles of Forecasting and Applications assignment at Monash University Malaysia? Many students find it challenging to apply ARIMA, ensemble, and bootstrap models while interpreting real-world economic data accurately. At Malaysia Assignment Help, our forecasting and econometrics experts deliver 100% plagiarism-free, AI-free reports that meet Monash university standards. Get professional support through our Online Assignment Writing Service in Malaysia and order your perfectly written forecasting assignment today!
