The question
A result can be computed correctly and still answer the wrong question. This reading sequence brings together data analytics, a familiar working tool, forecasting, and simulation validation. Read with one distinction in mind: reproducing a calculation tells us whether we can obtain the result again; deciding whether to use that result requires evidence about the problem it represents. Keep a record of the assumptions you encounter, who would have to accept them, and what observation would make you reconsider them.
A suggested reading sequence
How the chapters connect
- Chapter 1 · Modern Data Analytics for the Military Operational Researcher
- Use Hill’s distinctions among analytical approaches to state what the proposed analysis is supposed to establish.
- Chapter 2 · Microsoft Excel: The Universal Tool of Analysis
- Compare that purpose with the practical spreadsheet methods in Lindquist and Sulewski. A convenient implementation does not settle the suitability of the method.
- Chapter 13 · Wavelet Methods for Time Series Forecasting
- Nystrom and colleagues show why the structure and scale of a time series can matter to preprocessing and forecasting. Ask what a transformation preserves.
- Chapter 14 · Simulation Validation Literature Review
- Ledwith and colleagues address simulation validation, not time-series forecasting as such. Use their distinction between verification and validity to frame a separate question about evidence and intended use.
The connections and exercise are companion editorial material. Refer to the chapters for their full methods, evidence, and conclusions.
Questions to carry into the reading
- What decision is the proposed analysis supposed to inform? State it before choosing a method.
- Which assumptions come from the data, and which have been introduced by the analyst?
- What evidence would support using the result outside the conditions under which it was produced?
Try it in discussion
A small exercise
Choose an ordinary forecasting problem, such as demand for a campus service. Write a one-page plan identifying the available data, the quantity to predict, and a way to assess a forecast against observations that were not used to construct it. Do not calculate a forecast yet. Exchange plans with another reader and identify one assumption that would need testing before the proposed forecast could support a decision.
After the reading
Return to your original question. Identify which part of your answer now rests on a particular chapter, which part depends on your own assumptions, and which part remains unresolved. Give the chapter authors credit for the specific contribution you used. An answer can become more useful by stating a limit clearly; there is no need to make every chapter settle every part of the problem.