Predict
Estimating a value you do not have yet, or flagging that an observed value is out of the ordinary. This is the family that demands the most data and is easiest to get wrong, because a plausible prediction is hard to tell from a correct one. A simple method you can evaluate beats a clever one you cannot check.
What is going to happen, and what looks wrong?
2 entries
- Forecast next week’s salesKnow how much I will sell next week, from my own sales history.No generative AI needednégligeableInstantYour data never leaves
- Spot an anomaly in a metricGet paged when a production metric leaves its usual behaviour, and not the rest of the time.No AI neededFreeInstantYour data never leaves