Data scientist interview: how would you explain overfitting to a non-technical stakeholder?

Model answer

The skill being tested is communication, so avoid jargon. Use a simple analogy: a model that overfits is like a student who memorises last year's exam paper instead of understanding the subject — it scores perfectly on questions it has seen but fails on new ones. Translate that to the business: the model looks excellent on historical data but performs poorly on new, real-world cases, so its predictions cannot be trusted for decisions. Then, still plainly, explain why it happens — the model learned the noise and quirks of the training data rather than the true pattern, often because it was too complex or trained on too little data. Reassure the stakeholder that it is detectable and fixable: you check performance on data the model has not seen, and you address it with more data, a simpler model, or techniques that discourage memorising. Close by connecting it to their concern — you would rather ship a model that is reliably good on new data than one that merely looks perfect in testing. Keeping it concrete and honest builds trust with a non-technical audience.

This is a guide to structure and approach, not a script to memorise — say it in your own words.

Practice this yourself

Type your own answer below and get it scored instantly by the same AI that powers Vachix mock interviews.