polars-ops 0.46.0

More operations on Polars data structures
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

use polars_core::prelude::*;
use polars_core::with_match_physical_integer_polars_type;

use crate::series::ops::SeriesSealed;

fn log<T: PolarsNumericType>(ca: &ChunkedArray<T>, base: f64) -> Float64Chunked {
    ca.cast_and_apply_in_place(|v: f64| v.log(base))
}

fn log1p<T: PolarsNumericType>(ca: &ChunkedArray<T>) -> Float64Chunked {
    ca.cast_and_apply_in_place(|v: f64| v.ln_1p())
}

fn exp<T: PolarsNumericType>(ca: &ChunkedArray<T>) -> Float64Chunked {
    ca.cast_and_apply_in_place(|v: f64| v.exp())
}

pub trait LogSeries: SeriesSealed {
    /// Compute the logarithm to a given base
    fn log(&self, base: f64) -> Series {
        let s = self.as_series();
        if s.dtype().is_decimal() {
            return s.cast(&DataType::Float64).unwrap().log(base);
        }

        let s = s.to_physical_repr();
        let s = s.as_ref();

        use DataType::*;
        match s.dtype() {
            dt if dt.is_integer() => {
                with_match_physical_integer_polars_type!(s.dtype(), |$T| {
                    let ca: &ChunkedArray<$T> = s.as_ref().as_ref().as_ref();
                    log(ca, base).into_series()
                })
            },
            Float32 => s
                .f32()
                .unwrap()
                .apply_values(|v| v.log(base as f32))
                .into_series(),
            Float64 => s.f64().unwrap().apply_values(|v| v.log(base)).into_series(),
            _ => s.cast(&DataType::Float64).unwrap().log(base),
        }
    }

    /// Compute the natural logarithm of all elements plus one in the input array
    fn log1p(&self) -> Series {
        let s = self.as_series();
        if s.dtype().is_decimal() {
            return s.cast(&DataType::Float64).unwrap().log1p();
        }

        let s = s.to_physical_repr();
        let s = s.as_ref();

        use DataType::*;
        match s.dtype() {
            dt if dt.is_integer() => {
                with_match_physical_integer_polars_type!(s.dtype(), |$T| {
                    let ca: &ChunkedArray<$T> = s.as_ref().as_ref().as_ref();
                    log1p(ca).into_series()
                })
            },
            Float32 => s.f32().unwrap().apply_values(|v| v.ln_1p()).into_series(),
            Float64 => s.f64().unwrap().apply_values(|v| v.ln_1p()).into_series(),
            _ => s.cast(&DataType::Float64).unwrap().log1p(),
        }
    }

    /// Calculate the exponential of all elements in the input array.
    fn exp(&self) -> Series {
        let s = self.as_series();
        if s.dtype().is_decimal() {
            return s.cast(&DataType::Float64).unwrap().exp();
        }

        let s = s.to_physical_repr();
        let s = s.as_ref();

        use DataType::*;
        match s.dtype() {
            dt if dt.is_integer() => {
                with_match_physical_integer_polars_type!(s.dtype(), |$T| {
                    let ca: &ChunkedArray<$T> = s.as_ref().as_ref().as_ref();
                    exp(ca).into_series()
                })
            },
            Float32 => s.f32().unwrap().apply_values(|v| v.exp()).into_series(),
            Float64 => s.f64().unwrap().apply_values(|v| v.exp()).into_series(),
            _ => s.cast(&DataType::Float64).unwrap().exp(),
        }
    }

    /// Compute the entropy as `-sum(pk * log(pk)`.
    /// where `pk` are discrete probabilities.
    fn entropy(&self, base: f64, normalize: bool) -> PolarsResult<f64> {
        let s = self.as_series().to_physical_repr();
        polars_ensure!(s.dtype().is_primitive_numeric(), InvalidOperation: "expected numerical input for 'entropy'");
        // if there is only one value in the series, return 0.0 to prevent the
        // function from returning -0.0
        if s.len() == 1 {
            return Ok(0.0);
        }
        match s.dtype() {
            DataType::Float32 | DataType::Float64 => {
                let pk = s.as_ref();

                let pk = if normalize {
                    let sum = pk.sum_reduce().unwrap().into_series(PlSmallStr::EMPTY);

                    if sum.get(0).unwrap().extract::<f64>().unwrap() != 1.0 {
                        (pk / &sum)?
                    } else {
                        pk.clone()
                    }
                } else {
                    pk.clone()
                };

                let log_pk = pk.log(base);
                (&pk * &log_pk)?.sum::<f64>().map(|v| -v)
            },
            _ => s
                .cast(&DataType::Float64)
                .map(|s| s.entropy(base, normalize))?,
        }
    }
}

impl LogSeries for Series {}