Quantcast
Channel: Machine Learning
Viewing all articles
Browse latest Browse all 62716

Missing lots of data. Best way to complete my data set?

$
0
0

The data set I have is an old picture. It shows only the top 14 entries. For each key-value pair the value is a percentage. #1 entry's value is 0.62%. #14 entry's is 0.23%. All the rest is lost.

Intuitively, I know the distribution is very close to being uniform. I could just add the top rows percentages, subtract from 100, and divide by that many keys so they all have less or equal to 0.23. However, it would be better if I could find a special way to draw from a distribution as many times as needed, all draws should be less than .23 but should all add up to 100.

There must be a cool way to do this. Any ideas? TIA

submitted by suorm
[link] [6 comments]

Viewing all articles
Browse latest Browse all 62716

Trending Articles