False Positive Rate of 1:1.5M Sampling Process

  • Context: MHB 
  • Thread starter Thread starter karamand
  • Start date Start date
  • Tags Tags
    Positive Rate
Click For Summary

Discussion Overview

The discussion revolves around the probability of a false positive in a sampling process where items are classified as type A or type B. Participants explore the implications of a false positive rate of 1 in 1.5 million when a sample of 1 million yields one classification as type A. The conversation touches on the application of probability theory, particularly in the context of false positives.

Discussion Character

  • Exploratory, Technical explanation, Debate/contested, Mathematical reasoning

Main Points Raised

  • One participant states that the probability of a 'hit' being a false positive is approximately $\frac{1}{1.5\cdot 10^6} \approx 6.7 \cdot 10^{-5}$, but emphasizes that more information is needed to draw further conclusions about the other observations.
  • Another participant agrees with the initial calculation, suggesting that the sample size does not affect the probability of any individual 'hit' being a false positive.
  • A third participant introduces Bayes' Theorem, noting that it typically requires more information to analyze false positives but acknowledges that the given probability is already known and thus may not be necessary for this specific case.

Areas of Agreement / Disagreement

Participants generally agree on the calculation of the probability of a false positive for a single 'hit', but there is a lack of consensus on the necessity of additional information or methods, such as Bayes' Theorem, for a more comprehensive analysis.

Contextual Notes

Limitations include the absence of information about the actual distribution of type A and type B items in the population, which affects the overall probability assessment. The discussion does not resolve how to incorporate this information into the probability calculation.

karamand
Messages
6
Reaction score
0
I have a sampling process of a very large population in which all items are of type A or type B. I have an analysis of the sampled objects which classifies type A and gives the wrong identification (a false positive) 1 in 1.5 million times.
I take a sample of 1 million and find 1 'hit' i.e classified as type A. What is the probability that it is a false positive?
 
Physics news on Phys.org
philpq said:
I have a sampling process of a very large population in which all items are of type A or type B. I have an analysis of the sampled objects which classifies type A and gives the wrong identification (a false positive) 1 in 1.5 million times.
I take a sample of 1 million and find 1 'hit' i.e classified as type A. What is the probability that it is a false positive?

Hi philpq! Welcome to MHB! :)

Without more information, any 'hit' of type A has a probability of $\frac{1}{1.5\cdot 10^6} \approx 6.7 \cdot 10^{-5}$ of being a false positive.
We will still know basically nothing about the other 999999 observations without more information.
 
I like Serena said:
Hi philpq! Welcome to MHB! :)

Without more information, any 'hit' of type A has a probability of $\frac{1}{1.5\cdot 10^6} \approx 6.7 \cdot 10^{-5}$ of being a false positive.
We will still know basically nothing about the other 999999 observations without more information.

Thanks for your help. I suppose the answer is obvious when I think about it. The sample size is irrelevant. The probability of anyone 'hit' being a false positive is 1 in 1.5 million as stated :)
 
Usually questions about false positives use Bayes' Theorem and for that you need a lot more information.

$$P(+|\text{ (actually negative)})=\frac{P(\text{(actually negative)}|+) \cdot P(+)}{P(\text{actually negative})}$$

In the above, $+$ means "reads positive". However, you already have this probability so the above isn't necessary to calculate. I'm just pointing out that these topics are very often related. Here is an example "false positive" question you can read on Wikipedia.
 

Similar threads

  • · Replies 5 ·
Replies
5
Views
3K
  • · Replies 6 ·
Replies
6
Views
3K
  • · Replies 7 ·
Replies
7
Views
2K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 14 ·
Replies
14
Views
3K
  • · Replies 6 ·
Replies
6
Views
2K
  • · Replies 6 ·
Replies
6
Views
2K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 10 ·
Replies
10
Views
2K
  • · Replies 0 ·
Replies
0
Views
2K