openethereum/util/src/stats.rs

104 lines
3.4 KiB
Rust
Raw Normal View History

// Copyright 2015, 2016 Ethcore (UK) Ltd.
// This file is part of Parity.
// Parity is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
// Parity is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
// You should have received a copy of the GNU General Public License
// along with Parity. If not, see <http://www.gnu.org/licenses/>.
//! Statistical functions.
use bigint::uint::*;
/// Discretised histogram.
#[derive(Debug, PartialEq)]
pub struct Histogram {
/// Bounds of each bucket.
pub bucket_bounds: Vec<U256>,
/// Count within each bucket.
pub counts: Vec<u64>
}
impl Histogram {
/// Histogram of a sorted corpus if it at least spans the buckets. Bounds are left closed.
pub fn new(corpus: &[U256], bucket_number: usize) -> Option<Histogram> {
if corpus.len() < 1 { return None; }
let corpus_end = corpus.last().expect("there is at least 1 element; qed").clone();
let corpus_start = corpus.first().expect("there is at least 1 element; qed").clone();
// Bucket needs to be at least 1 wide.
let bucket_size = {
// Round up to get the entire corpus included.
let raw_bucket_size = (corpus_end - corpus_start + bucket_number.into()) / bucket_number.into();
if raw_bucket_size == 0.into() { 1.into() } else { raw_bucket_size }
};
let mut bucket_end = corpus_start + bucket_size;
let mut bucket_bounds = vec![corpus_start; bucket_number + 1];
let mut counts = vec![0; bucket_number];
let mut corpus_i = 0;
// Go through the corpus adding to buckets.
for bucket in 0..bucket_number {
while corpus.get(corpus_i).map_or(false, |v| v < &bucket_end) {
// Initialized to size bucket_number above; iterates up to bucket_number; qed
counts[bucket] += 1;
corpus_i += 1;
}
// Initialized to size bucket_number + 1 above; iterates up to bucket_number; subscript is in range; qed
bucket_bounds[bucket + 1] = bucket_end;
bucket_end = bucket_end + bucket_size;
}
Some(Histogram { bucket_bounds: bucket_bounds, counts: counts })
}
}
#[cfg(test)]
mod tests {
use bigint::uint::U256;
use super::Histogram;
#[test]
fn check_histogram() {
2016-11-28 13:20:49 +01:00
let hist = Histogram::new(slice_into![643,689,1408,2000,2296,2512,4250,4320,4842,4958,5804,6065,6098,6354,7002,7145,7845,8589,8593,8895], 5).unwrap();
let correct_bounds: Vec<U256> = vec_into![643, 2294, 3945, 5596, 7247, 8898];
assert_eq!(Histogram { bucket_bounds: correct_bounds, counts: vec![4,2,4,6,4] }, hist);
}
#[test]
fn smaller_data_range_than_bucket_range() {
assert_eq!(
2016-11-28 13:20:49 +01:00
Histogram::new(slice_into![1, 2, 2], 3),
Some(Histogram { bucket_bounds: vec_into![1, 2, 3, 4], counts: vec![1, 2, 0] })
);
}
#[test]
fn data_range_is_not_multiple_of_bucket_range() {
assert_eq!(
2016-11-28 13:20:49 +01:00
Histogram::new(slice_into![1, 2, 5], 2),
Some(Histogram { bucket_bounds: vec_into![1, 4, 7], counts: vec![2, 1] })
);
}
#[test]
fn data_range_is_multiple_of_bucket_range() {
assert_eq!(
2016-11-28 13:20:49 +01:00
Histogram::new(slice_into![1, 2, 6], 2),
Some(Histogram { bucket_bounds: vec_into![1, 4, 7], counts: vec![2, 1] })
);
}
2016-11-03 08:22:35 +01:00
#[test]
fn none_when_too_few_data() {
2016-11-28 13:20:49 +01:00
assert!(Histogram::new(slice_into![], 1).is_none());
2016-11-03 08:22:35 +01:00
}
}