Time Series Prediction Project Two choices
Use one of the US Census time series linked. here (note that the Census will allow you to check “seasonally adjusted” or “not seasonally adjusted.” You’ll almost certainly want “not seasonally adjusted” since it will help you to show that you can deal with seasonality).
Find another time series of interest to you on the internet (any time series OK, but be sure to include a link to the data.)
- Perform a time series analysis.
- Consider is it horizontal? Does it show a trend? Seasonality?
- Consider smoothing (moving average or exponential smoothing).
- Include graphs of both.
- Do you believe they improve the model? Why or why not?
5 points, subjective: was this interesting? (interesting data, interesting conclusion, interesting method)
5: very 4: largely 3: somewhat 2: a little 1: no 0: no data
5 points, graphs:
5 attractive 4 good 3 passable 2 OK 1 cringeworthy 0 no graph
20 points, Time series tools considered (horizontal; seasonality; trend; cycle)
For each of the four tools above:
5 – considered and clearly explained
4 – considered if somewhat unclear
3 – considered with no explanation
2 – possibly considered
1 – no mention of that tool
0 – none considered
10 points: Smoothing and moving average
5 – judgment given, graphs support
4 – judgment given but graphs don’t support, or good graphs but no clear judgment
3 – no clear judgment and graphs don’t make the argument clearly on their own
2 – no graphs
1 – no clear judgment and no graphs
0 – smoothing not considered.