MATLAB数据预测

EMET3007/8012 – Forecasting Project
For this project you must choose a monthly or quarterly time series, make a fore- cast, and evaluate your success. This means you will need to turn in three reports.
Project Description
The Project Description report will be due on the 20th of September. This report is a brief description (1-2 paragraphs) of the variable you intend to forecast. You must include the following:
• A precise description of the variable you will be forecasting
• The original source for the time-series
• Where you obtained the series
• Which date you intend to forecast, when that data will be released, and where the data will be released.
If you have access to proprietary data and wish to use that, that’s fine1. Otherwise you can use any data series you wish. E.g. US data series are fine to use if you want. The Project Description should not be more than one page.
Very Important: The time-series you are forecasting must have a new real- ization some time between the 28th of October and the 8th of November.
Project Report
This is the main forecast report. It is due no later than Thursday the 7th of November (but may be due earlier). This report will include
• A description of the data series, historical data, and time-series problems of the series
1Keep in mind that you will need to submit the data with your main report
1

• A description of a number of forecasting methods which you have used on this data
• An evaluation of the efficacy of the forecasting models used
• An indication of the optimal forecasting method
• Forecasts of the series for the coming year using the optimal method
• This should include point forecasts as well as more developed forecasts
• Anything else you think would be of interest
All code used must be included as an appendix. Use of appropriate graphs is strongly recommended. The report should be approximately 10 pages long, ex- cluding appendices.
Due Date: This report is due the day before the new realization of the data series. This means it will be due no later than the 7th of November, and may be due as early as the 27th of October. This is so you can evaluate your forecast.
Forecast Evaluation
This is a short evaluation of you forecast, where you compare your one-step-ahead forecast with the actual realization. Would a decision maker have been wise to follow your forecast? This is due on the 8th of November. The evaluation should be no longer than half a page, and could be shorter.
2

Marking Guide
You will be graded based on a combination of the following factors:
• Quality of Expression: How well the report is written and expressed. This is primarily concerned with the flow and presentation of the report.
• Quality of Analysis: The forecasting analysis should be well-developed and argued. All analysis should be free from errors. Appropriate regressors should be identified and utilised. A range of appropriate models should be attempted, and the best model identified. More difficult to implement models should be included to demonstrate econometric sophistication.
• Topic Choice: The relevance and importance of the topic.
• Use of Appropriate References: The quality and relevance of references
used.
• Referencing: References must be provided, the referencing system used must be consistent, and of sufficient detail to be able to find the references. All data used used must be sourced.
• Accuracy of Forecast: The actual accuracy of the forecast is assessed. In cases where the forecast is inaccurate, the quality of argument as to why it was inaccurate is also assessed.
3