cortical-io 0.1.14

Cortical.io API client
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143

use std::ops::{Add, Div, Mul};

use serde::{Deserialize, Serialize};

use crate::Fingerprint;

#[derive(Default, Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct FingerprintSimilarity {
    pub exp_vec_left: Vec<f64>,
    pub exp_vec_right: Vec<f64>,
}

impl FingerprintSimilarity {
    pub fn new(left: &Fingerprint, right: &Fingerprint) -> Self {
        let exp_vec_left = left.expand_t::<f64>(128 * 128);
        let exp_vec_right = right.expand_t::<f64>(128 * 128);

        Self {
            exp_vec_left,
            exp_vec_right,
        }
    }

    #[inline(always)]
    pub fn euclidean_distance(&self) -> f64 {
        self.exp_vec_left
            .iter()
            .zip(self.exp_vec_right.iter())
            .map(|(l, r)| (l - r).powi(2))
            .sum::<f64>()
            .sqrt()
    }

    #[inline(always)]
    pub fn jaccard_index(&self) -> f64 {
        self.exp_vec_left
            .iter()
            .zip(self.exp_vec_right.iter())
            .map(|(l, r)| (l - r).abs())
            .sum::<f64>()
            .div(2.0)
    }

    // normalized euclidean distance
    // 0.5*((norm((x-mean(x))-(y-mean(y)))^2)/(normalize(x-mean(x))^2+normalize(y-mean(y))^2))
    #[inline(always)]
    pub fn normalized_euclidean_distance(&self) -> f64 {
        let left_mean =
            self.exp_vec_left
                .iter()
                .sum::<f64>()
                .div(
                    self.exp_vec_left
                        .len() as f64,
                );

        let right_mean =
            self.exp_vec_right
                .iter()
                .sum::<f64>()
                .div(
                    self.exp_vec_right
                        .len() as f64,
                );

        let (left_norm, right_norm, sum) =
            self.exp_vec_left
                .iter()
                .zip(self.exp_vec_right.iter())
                .fold(
                    (0.0, 0.0, 0.0),
                    |(left_norm, right_norm, sum), (l, r)| {
                        (
                            left_norm.add((l - left_mean).powi(2)),
                            right_norm.add((r - right_mean).powi(2)),
                            sum.add((l - left_mean - (r - right_mean)).powi(2)),
                        )
                    },
                );

        sum / (left_norm + right_norm)
    }

    #[inline(always)]
    pub fn normalized_euclidean_similarity(&self) -> f64 {
        1.0 - self.normalized_euclidean_distance()
    }

    #[inline(always)]
    pub fn cosine_similarity(&self) -> f64 {
        let mut sum = 0.0;
        let mut left_sum = 0.0;
        let mut right_sum = 0.0;

        for i in 0..self.exp_vec_left.len() {
            sum += self.exp_vec_left[i] * self.exp_vec_right[i];
            left_sum += self.exp_vec_left[i] * self.exp_vec_left[i];
            right_sum += self.exp_vec_right[i] * self.exp_vec_right[i];
        }

        (sum / (left_sum * right_sum).sqrt()).add(1.0).div(2.0)
    }

    #[inline(always)]
    pub fn overlapping_all(&self) -> u64 {
        let left_hs = &self.exp_vec_left;
        let right_hs = &self.exp_vec_right;

        left_hs
            .iter()
            .zip(right_hs.iter())
            .filter(|(l, r)| l.eq(r) && (**l).eq(&1.0))
            .count() as u64
    }

    #[inline(always)]
    pub fn weighted_scoring(
        &self,
    ) -> f64 {
        self.cosine_similarity()
            .mul(self.normalized_euclidean_similarity())
            .sqrt()
    }

    #[inline(always)]
    pub fn pearson_r_coeff(&self) -> f64 {
        let left_mean = self.exp_vec_left.iter().sum::<f64>() / self.exp_vec_left.len() as f64;
        let right_mean = self.exp_vec_right.iter().sum::<f64>() / self.exp_vec_right.len() as f64;

        let (r1, r2) =
            self.exp_vec_left
                .iter()
                .zip(self.exp_vec_right.iter())
                .fold((0.0, 0.0), |(_s1, s2), (l, r)| {
                    (
                        s2.add((l - left_mean).mul(r - right_mean)),
                        s2.add((l - left_mean).powi(2).mul((r - right_mean).powi(2))),
                    )
                });

        r1.div(r2.sqrt())
    }
}