Tuesday, June 07, 2016

[dtloizxg] Estimating probabilities from seemingly unreliable data

Given noisy empirical data about an event, estimate the probability or frequency of its future occurrence.  A difficult task for natural or artificial intelligence.  The nature of the noise could be extremely complicated.  It takes intelligence to understand the nature of the noise and compensate for it.

But a predictive algorithm can later have its accuracy evaluated as the future rolls around.

Inspired by rare side effects of osteoporosis medication: do the benefits outweigh the risks?  One needs to know probabilities.

No comments :