bentov

1D histogram sketching
README

An OCaml implementation of histogram-sketching algorithm described in
A Streaming Parallel Decision Tree
Algorithm

by Yael Ben-Haim and Elad Tom-Tov . Included is a command-line
utility bt, which can read a file (or stdin) containing numbers, one
per line, and output a representation of the approximated
distribution.

For example, to approximate 10 quantiles of 1M data in U(0,1):

echo "" | awk '{ for ( i=0 ; i < 1e6 ; i++ ) { print rand() } }' | bt -n 20 -u 10

In this example, the size of the approximating histogram is 20. For
additional details, bt --help .

To install:

opam install bentov
Install
Published
24 Jun 2020
Maintainers
Sources
bentov-1.tbz
sha256=9dd08f88b554ee5d8127f683cd4530773763091d692c86b7ddd4794a0fbad8e3
sha512=622efa18f1a5a3c2968953570ac30d37d9871e011c7d340cc11d3ced35ee2f8327e8bd1016e964e15536a4ff0c500a5b76ebb525d35a52cc81d03b997f45464f
Dependencies
ocaml
>= "4.08.0"
cmdliner
>= "1.0.4"
dune
>= "2.5"
Reverse Dependencies