Skip to content

Navigation: <-- Working as a Team on a DS Project | Part Index | Main Index | What a Finished Project Looks Like -->

Reality-Checking Your Idea

Requires: Working as a Team on a DS Project

Motivation: You or your group have committed to a question. Before investing further in it, it is worth spending ten minutes checking whether it is actually workable. Not perfect, just doable.

This nugget walks you through two quick checks: data feasibility and scope. They will help you distinguish a viable idea from one that needs adjusting, without overthinking it.

Table of Contents

Data Feasibility

Ask one practical question first:

Can you actually access relevant data in the near term?

Not: is the data perfect? Not: is the dataset large enough? Just: is there something you can work with?

A few things worth checking:

  • Is data already recorded and accessible: a sensor log, a spreadsheet, a database?
  • If not, can you collect or record it yourself within a week (or a few)?
  • If the data belongs to an institute or company, do you have permission to use it?

Almost any form of data works: tabular measurements, time series, images from a camera or microscope, or a spreadsheet you fill in during your regular lab work. If you already have something sitting around, you are ahead.

In terms of 🖝 CRISP-DM, you will recognize this as a light pass at the Data Understanding phase. Not yet a full analysis, just a check that starting is worth it.

If a candidate idea falls apart here, that is useful information. Adjust the question or pick the next candidate.

Scope and Timeline

The other thing worth checking is whether the idea fits inside a semester alongside everything else in the course.

A few rough questions:

  • Can you imagine having a result to show within a few weeks of focused work?
  • Can you say in one sentence what "done" would look like?
  • Does the idea depend on data that does not exist yet, or on techniques you would need months to learn?

You do not need exact answers. The point is to catch ideas that are really little research programmes in disguise. That's what curiosity is about! Just notice when an idea is within reach.

If the scope seems too large, the fix is usually to narrow the question, not to abandon the topic. "Understand everything about vibration in our system" becomes "detect one specific fault type from accelerometer readings."

Summary

  • The key feasibility question is simple: can you access relevant data in the near term?
  • Almost any data form works. Something you already have is a strong starting point.
  • Check that the idea is completable in a semester and that you can sketch what "done" looks like.
  • If an idea falls short, narrow the question before abandoning the topic.

As always: Happy learning, happy life! 🫶

Navigation: <-- Working as a Team on a DS Project | Part Index | Main Index | What a Finished Project Looks Like -->

Script v1.4 (2026-06-10) · FGN