## Forecasting

Skill: The purpose of this assignment is to help you practice the following skills that are essential
to your success in this course and in professional life beyond school:
Formulate a problem statement related to a forecasting question.
Design a strategy to answer the forecasting question.
Collect, clean, and use publicly available time-series data.
Evaluate various forecasting models to get the best possible forecast.
Select the best model based on model selection criteria and accuracy measures.
Execute appropriate statistical tests to obtain better forecasting accuracy.
Interpret, communicate, and present forecast results in a report.
Knowledge: This assignment will also help you to become familiar with the following valuable
content knowledge in this discipline.
Defining a forecasting problem.
Utilizing a statistical program to compute, visualize, and analyze time-series data in economics, business, and the social sciences.
Performing exploratory analysis.
Selecting an appropriate statistical model among alternative models.
Validating the selected statistical model.
Interpreting models using parameters.
Forecasting based on the selected statistical model.
Assessing the accuracy of forecasts.
Interpreting and communicate results effectively.
Task: The final project will consist of a brief report between 10-12 double-spaced pages, including
relevant tables and figures. To begin, you may choose your time-series dataset. Choose a dataset
that has a time component and a variable to analyze over time. Choose a dataset that you would
like to analyze according to the techniques that are outlined in the textbook and were discussed
in class. Find the model that you believe fits your data best and build a forecast from this model.
Assess the validity of that forecast. You must follow the five basic steps in a forecasting task
outlined in section 1.6 of your textbook.
1
Your report should be structured as follows:
1. Abstract (no more than 250 words): A summary of your basic findings.
2. Introduction (1-3 pages): A brief introduction/motivation to the problem at hand, relevant
details about the data, additional relevant scientific information from searching the web, for
example, and what is to be addressed.
3. Data Description (at most one page): A brief introduction to the data, data sources, and
variable definitions.
4. Statistical Methods (1-2 pages): A discussion and justification of the methods you have used
to analyze the data, and how you went about analyzing the data. Dont forget to describe in
some detail how and why the particular model was selected.
5. Results (2-3 pages): A presentation of the results of your analysis. Interpretations should
include a discussion of statistical versus the practical importance of the results.
6. Discussion (1-2 pages): A synopsis of your findings and any limitations your study may suffer
from. Present conclusions in terms that non-statisticians will understand. Quantitative and
qualitative aspects should be discussed.
Your report should be brief and to the point! It should be written in a language that is
understandable to the scientific community.
Criteria for Success:
1. Explanation: Excellent reports will use a data set a student is interested in and apply domain
knowledge that describes why the forecasting is useful or important. This importance will be
explained clearly in the introduction and conclusion of the report and will be evident to the
reader. Reports should be clear and concise and use scientific language.
2. Process-oriented: Good reports will also follow the 5 basic steps in a forecasting task described
in the textbook. Be sure to apply all the tools used and models discussed this semester to
determine the best forecast, and clearly show why this forecast is the best using the accuracy
measures. Reports should follow the task steps described above in the correct order.
3. Evidence-based: Use accuracy measures to back up the selected model. Show why alternative
models are not as efficient with your specific dataset.