4 ms·
These articles always seem to imply that you can use those traits for detection. Can you? I mean, the numbers and wording imply that you can, but some of the st
by romaniv 12y ago
These articles always seem to imply that you can use those traits for detection. Can you? I mean, the numbers and wording imply that you can, but some of the stats are not very clear or intuitive.
For example, just because group X usually doesn't eat lunch doesn't mean that not eating lunch is a good trait to detect them in the general population.
Also, 6% of outlook.com is used for fraud? This is a huge percentage.
How does this company detect multiple accounts on the device?
- lilsunnybee 12y agoI think they mean 6% of the email addresses used for fraudulent transactions were outlook.com addresses, at least in the data they were analyzing. 6% of fraud perpetrators use outlook.com is a much different statistic than 6% of outlook.com users commit fraud.
- roel_v 12y agoOnly only 'you can', it is being done by thousands of companies across the world every second. (not these properties directly, just building a statistical model of fraudulent users and comparing transactions to that model to flag potentially fraudulent transactions - in wire transfers, online purchases, credit card use, ...)