deepcorr_normalization/
min_max.rs

1/*
2This file is part of DeepCorr.
3
4DeepCorr is free software: you can redistribute it and/or modify it under 
5the terms of the GNU General Public License as published by the Free 
6Software Foundation, either version 3 of the License, or any later version.
7
8DeepCorr is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; 
9without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR 
10PURPOSE. See the GNU General Public License for more details.
11
12You should have received a copy of the GNU General Public License along with 
13DeepCorr. If not, see <https://www.gnu.org/licenses/>.
14*/
15
16use ndarray::{Array2, Axis};
17use crate::NormError;
18
19pub struct MinMaxNormalizer {
20    pub epsilon: f64, 
21}
22
23impl Default for MinMaxNormalizer {
24    fn default() -> Self {
25        Self { epsilon: 1e-10 }
26    }
27}
28
29impl MinMaxNormalizer {
30    pub fn new(epsilon: f64) -> Self {
31        Self { epsilon }
32    }
33
34    pub fn normalize(&self, data: &Array2<f64>) -> Result<Array2<f64>, NormError> {
35        if data.is_empty() {
36            return Err(NormError::EmptyInput);
37        }
38
39        let min = data.fold_axis(Axis(0), f64::INFINITY, |&a, &b| f64::min(a, b));
40        let max = data.fold_axis(Axis(0), f64::NEG_INFINITY, |&a, &b| f64::max(a, b));
41
42        let mut normalized = data.clone();
43
44        for (i, mut col) in normalized.axis_iter_mut(Axis(1)).enumerate() {
45            let col_min = min[i];
46            let col_max = max[i];
47            let range = col_max - col_min;
48
49            col.mapv_inplace(|x| (x - col_min) / (range + self.epsilon));
50        }
51
52        Ok(normalized)
53    }
54}
deepcorr_normalization/min_max.rs

deepcorr_normalization/
min_max.rs
