Overview
The goal of this project is for you to complete a time series analysis using the forecasting methods discussed in the lessons and text. You will choose one of the time series provided from the US Census time series website. (Links to an external site.)
Steps and Prep
For best understanding please review the details of the project in conjunction with Week 8 – Lessons and Video Lecture. Warning — you may not use the same retail sales data I chose to use in the example.
Go to the US Census Time Series Dashboard (Links to an external site.). Choose a variable. Take a quick look at the variable with the census plot tool. You can decide which variable to download for analysis and the time period.
Download the data as a TXT file. Either open right away or save the file and open with excel.
How to get the data:
Clean the data.
If necessary manipulate the data so that it is an appropriate format for time series analysis.
Create an appropriate date variable.
For regression create t variable and seasonal dummy variables.
Explore methods
Plot the data to explore patterns and to determine an appropriate forecasting method.
Use XLSTAT to forecast the next time period using at least two methods: moving average, exponential smoothing and/or regression.
Calculate the errors MSE, MAPE and a future value of each forecasting method, and decide which method is the best overall based on the lowest error.
Build the Best Forecast
Defend your forecasting method.
Plot the forecasted model over your data.
Provide a write up with answering the questions below.
Requirements and Scoring
Explore the time series
(5 points) Plot the time series. Paste plot. Edit axis labels so they are clean and understood by someone who is not familiar with the data.
(5 points) Describe the time series. Be sure to include the variable names and any relevant contextual information.
Does it show any specific patterns? I.e. horizontal, seasonal, trend, cyclical.
What is the variable and what might cause such patterns?
Are there any abrupt changes in the patterns? If so, when do they happen? Are they just an outlier or are they an overall shift? Are there any known historical events around that time period that might help explain the abrupt change? Consider whether a trend continues after the change and whether time periods before the change should be included in a forecast for future values.
Choose Forecast Methods
(2 points) Based on the patterns in your data, choose two methods to forecast your series. State the methods to which you chose to forecast your data.
(4 points) Explain why these methods are appropriate based on the pattern of the data.
(4 points) Based on your chosen methods, provide the following values and why or how you obtained them:
k value (for moving average)
smoothing constant (exponential smoothing)
and/or the regression equation
(10 points) For both methods, construct a plot with the forecasted model over the original data.
Edit axis labels so they are clean and understood by someone who is not familiar with the data.
Identify the forecasting method depicted in the plot.
Paste plots.
(6 points) Provide a table with your chosen methods. Include the MSE and MAPE and next time periods predicted value.
Your “Best” Method
(3 points) Determine which method is the best at predicting your future value. Explain why it is the “best” at forecasting your series.
(3 points) In a sentence state your forecasted value in context.
(3 points) Beyond the lowest MSE or MAPE, which method did you prefer? Why? Would you recommend choosing a different method even if it has a higher error? Why?
(5 points) Logistics – Include a Title page. Check spelling and grammar. Neatness – Provide responses in logical order. Plots are pasted inline with text. Use a simple and common font.
Peer Reviews and Deadlines
All due dates in Week 8 by 11:59PM PST
Original submission made by Wednesday
If you miss the original submission due date you will not be eligible for peer reviews nor will you be able to review your peers. You may still submit for instructor grading by Sunday.
Peer Reviews
At least 1 peer will “grade” and provide feedback on your work by the end of the day on Friday. Review the feedback, make adjustments.
You will do the same for at least 2 peers.
Resubmit by Sunday.
If you choose not to resubmit your work, the original submission will be graded by the instructor within 1 week.
Rubric Scoring
Scores are based on the point values in the instructions and rubric below.
Optional Group Submission
If you chose to complete this project in a group, then one person in your group will need to submit to the group submission option by Sunday at 11:59 PM (Next tab).
The group aspect eliminates the requirement for a peer review. You will earn the 20 points from the peer review by choosing the group option.
It is expected that you worked together and evaluated each others work. In order to so this option you need to have informed me by Wednesday of Week 7 in the Group Project Survey.
Last Completed Projects
| topic title | academic level | Writer | delivered |
|---|