Reference no: EM133529998

This project you are to provide a written exposition that thoroughly describes the complete analysis of a dataset. The aim of this project to perform forecasting, which is an important pan in time series analysis. It is important to think about the Box-Jenkins procedures whereby you must show the essential steps in setting up a certain forecasting model, which are:

1. Model identification (Chapter 6)
Examine the data to see which member of the class of ARIMA processes appears to be most appropriate.

2. Parameter estimation; i.e., model fitting (Chapter 7)
Estimate the parameters of the chosen model as described in Chapter 7.

3. Model diagnostics and checking (Chapter 8)
Examine the residuals from the fitted model to see if it is adequate as you have learned in Chapter 8 (possibly revisit the model specification phase at this point, based on what you learn from the diagnostics).

4. Consideration of alternative models if necessary (Chapter 8 & 10)
If the first model appears to be inadequate for some reason, then alternative ARIMA models may be tried until a satisfactory model is found.

4. Forecasting (Chapter 9 & 10)
After you select the best model, you have to provide the forecast of the values for the selected series at future times. You should provide point estimate and a 95% prediction interval of the predicted values. Check the adequacy of the fitted forecast model through examining whether the residuals have any undesirable properties such as significant autocorrelation and non-normality.

Recommended guidelines for choosing a dataset
- Use other sources (e.g., datasets online datasets such as those available from World Bank, OECD, Yahoo finance, Palestinian Central Bureau of Statistics (PCBS), etc.) to find an interesting data set. The more interesting the better!
- The responses A (i.e., the time series variable) should be continuous in nature. Also, make sure you completely understand the sampling frequency; e.g., are the data collected on annual, quarterly, monthly, weekly, or daily basis.
- Choose a dataset in an area you are interested in! You should be able to demonstrate a working knowledge of the subjecl area.
- Don't choose a dataset that is very small (e.g., let's stay away from series where n < 50). Ideally, we want n » 75 or so, but this is just a guideline. Remember that many statistical methods we have discussed exploit asymptotic distribution theory, so we want to apply these methods to suitably lengthy data sets.
- Remember that you have to obtain my approval to confirm your selected dataset before you begin your project.

Suggestion for forecasting
It might be a good idea to "withhold" some of the data from your series towards the end of it so that you can compare your forecasts to the actual values of the process. For example, suppose that you have a series of length n -- 120. Perform the specification, fit, and diagnostics on the first 115 observations and withhold the last 5. Then, when you forecast, you can compare your first 5 forecasts to the actual last 5 observations-this will give you an idea on how accurate/precise your forecasting is. Hopefully, your forecasts are "close" to the actual values in the series!

Outline of the written project (in this order)

1) Title page and abstract. You must prepare a title page with an appropriate title and abstract. The abstract should go on the title page. An abstract is a very high-level written summary of the entire project. Main points and findings only. The abstract should not exceed 150 words.

2) Introduction. In this part you are to introduce the reader to the dataset and to the area to which it pertains. For example, if you are analyzing the data on unemployment rate (UNR) or Consumer Price Index (CPI) in Palestine, you should describe why this is an important problem to investigate and give the reader a review of pertinent background information about unemployment or CPI in Palestine. This should be written at a very basic level (i.e., no mathematics or notation). Remember the cJient reader may not know anything about the area in which you are writing. This part should be written in at least 1 page with appropriate citation.

3) Model specification. This is the backbone of the project and will be the longest in length. In this section, you want to describe, in clear detail, the data analysis used to specify your candidate models. Avoid writing things like, "I tried this, and then I tried that
", this is not an academic language.

4) Fitting and Diagnostics. This part of the project should describe the model fitting and diagnostics lechniques you used, yvith the goal of identifying a "final" model for forecasting. Identify also what possible deficiencies your final model has. Remember, no model is perfect.

5) Forecasting. This section is the merit of your project and should describe the techniques you used to forecast future values (see "Suggestion for forecasting" on the previous page). Why is forecasting important? What impacts could your forecasting have?

6) Conclusion. Here you are to offer a summary of what you did in the project and draw