stats/
online.rs

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
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
use std::default::Default;
use std::fmt;
use std::iter::{FromIterator, IntoIterator};

use num_traits::ToPrimitive;

use Commute;

/// Compute the standard deviation of a stream in constant space.
pub fn stddev<I>(it: I) -> f64
        where I: Iterator, <I as Iterator>::Item: ToPrimitive {
    it.collect::<OnlineStats>().stddev()
}

/// Compute the variance of a stream in constant space.
pub fn variance<I>(it: I) -> f64
        where I: Iterator, <I as Iterator>::Item: ToPrimitive {
    it.collect::<OnlineStats>().variance()
}

/// Compute the mean of a stream in constant space.
pub fn mean<I>(it: I) -> f64
        where I: Iterator, <I as Iterator>::Item: ToPrimitive {
    it.collect::<OnlineStats>().mean()
}

/// Online state for computing mean, variance and standard deviation.
#[derive(Clone, Copy)]
pub struct OnlineStats {
    size: u64,
    mean: f64,
    variance: f64,
}

impl OnlineStats {
    /// Create initial state.
    ///
    /// Population size, variance and mean are set to `0`.
    pub fn new() -> OnlineStats {
        Default::default()
    }

    /// Initializes variance from a sample.
    pub fn from_slice<T: ToPrimitive>(samples: &[T]) -> OnlineStats {
        samples.iter().map(|n| n.to_f64().unwrap()).collect()
    }

    /// Return the current mean.
    pub fn mean(&self) -> f64 {
        self.mean
    }

    /// Return the current standard deviation.
    pub fn stddev(&self) -> f64 {
        self.variance.sqrt()
    }

    /// Return the current variance.
    pub fn variance(&self) -> f64 {
        self.variance
    }

    /// Add a new sample.
    pub fn add<T: ToPrimitive>(&mut self, sample: T) {
        let sample = sample.to_f64().unwrap();
        // Taken from: http://goo.gl/JKeqvj
        // See also: http://goo.gl/qTtI3V
        let oldmean = self.mean;
        let prevq = self.variance * (self.size as f64);

        self.size += 1;
        self.mean += (sample - oldmean) / (self.size as f64);
        self.variance = (prevq + (sample - oldmean) * (sample - self.mean))
                        / (self.size as f64);
    }

    /// Add a new NULL value to the population.
    ///
    /// This increases the population size by `1`.
    pub fn add_null(&mut self) {
        self.add(0usize);
    }

    /// Returns the number of data points.
    pub fn len(&self) -> usize {
        self.size as usize
    }
}

impl Commute for OnlineStats {
    fn merge(&mut self, v: OnlineStats) {
        // Taken from: http://goo.gl/iODi28
        let (s1, s2) = (self.size as f64, v.size as f64);
        let meandiffsq = (self.mean - v.mean) * (self.mean - v.mean);
        let mean = ((s1 * self.mean) + (s2 * v.mean)) / (s1 + s2);
        let var = (((s1 * self.variance) + (s2 * v.variance))
                   / (s1 + s2))
                  +
                  ((s1 * s2 * meandiffsq) / ((s1 + s2) * (s1 + s2)));
        self.size += v.size;
        self.mean = mean;
        self.variance = var;
    }
}

impl Default for OnlineStats {
    fn default() -> OnlineStats {
        OnlineStats {
            size: 0,
            mean: 0.0,
            variance: 0.0,
        }
    }
}

impl fmt::Debug for OnlineStats {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        write!(f, "{:.10} +/- {:.10}", self.mean(), self.stddev())
    }
}

impl<T: ToPrimitive> FromIterator<T> for OnlineStats {
    fn from_iter<I: IntoIterator<Item=T>>(it: I) -> OnlineStats {
        let mut v = OnlineStats::new();
        v.extend(it);
        v
    }
}

impl<T: ToPrimitive> Extend<T> for OnlineStats {
    fn extend<I: IntoIterator<Item=T>>(&mut self, it: I) {
        for sample in it {
            self.add(sample)
        }
    }
}

#[cfg(test)]
mod test {
    use {Commute, merge_all};
    use super::OnlineStats;

    #[test]
    fn stddev() {
        // TODO: Convert this to a quickcheck test.
        let expected = OnlineStats::from_slice(&[1usize, 2, 3, 2, 4, 6]);

        let var1 = OnlineStats::from_slice(&[1usize, 2, 3]);
        let var2 = OnlineStats::from_slice(&[2usize, 4, 6]);
        let mut got = var1;
        got.merge(var2);
        assert_eq!(expected.stddev(), got.stddev());
    }

    #[test]
    fn stddev_many() {
        // TODO: Convert this to a quickcheck test.
        let expected = OnlineStats::from_slice(
            &[1usize, 2, 3, 2, 4, 6, 3, 6, 9]);

        let vars = vec![
            OnlineStats::from_slice(&[1usize, 2, 3]),
            OnlineStats::from_slice(&[2usize, 4, 6]),
            OnlineStats::from_slice(&[3usize, 6, 9]),
        ];
        assert_eq!(expected.stddev(),
                   merge_all(vars.into_iter()).unwrap().stddev());
    }
}