An estimate can help you judge the size of a problem before collecting every detail. The useful part is a transparent chain of assumptions, rather than a final number that appears more certain than its inputs.
Break the question into quantities you can approximately describe. Record the units and use ranges where a value is unclear. Then consider which assumption has the largest influence on the result.
Use the estimate to choose the next observation. A rough answer may show that one missing detail deserves attention while several others barely affect the decision. An estimate becomes a thinking tool when another person can see how to question and improve it.
An example to consider.
Consider estimating a reading task using an assumed pace. The useful result includes the assumption, so another reader can adapt it to their own pace.
Put it in perspective.
A model chooses which features to keep. Ask which assumptions make its result useful and which real conditions sit outside its description.
- Write the assumptions beside the calculation.
- Keep units consistent.
- Identify the input that most changes the result.
Follow a related question
Choose one question the trial can answer.
Try a big idea on a small scaleDescribe accepted input and expected output.
An interface is a promiseKeep learning
Related background to continue exploring this subject.
NIST: the International System of Units NIST: statistical engineering
