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
170
171
172
173
174
175
176
177
178
179
180
181
182
|
#pragma once
#include <AggregateFunctions/IAggregateFunction.h>
#include <Columns/ColumnVector.h>
#include <Columns/ColumnTuple.h>
#include <Common/assert_cast.h>
#include <DataTypes/DataTypeNullable.h>
#include <DataTypes/DataTypesNumber.h>
#include <DataTypes/DataTypeTuple.h>
#include <IO/ReadHelpers.h>
#include <IO/WriteHelpers.h>
#include <limits>
namespace DB
{
struct Settings;
namespace ErrorCodes
{
}
template <typename T>
struct AggregateFunctionSimpleLinearRegressionData final
{
size_t count = 0;
T sum_x = 0;
T sum_y = 0;
T sum_xx = 0;
T sum_xy = 0;
void add(T x, T y)
{
count += 1;
sum_x += x;
sum_y += y;
sum_xx += x * x;
sum_xy += x * y;
}
void merge(const AggregateFunctionSimpleLinearRegressionData & other)
{
count += other.count;
sum_x += other.sum_x;
sum_y += other.sum_y;
sum_xx += other.sum_xx;
sum_xy += other.sum_xy;
}
void serialize(WriteBuffer & buf) const
{
writeBinary(count, buf);
writeBinary(sum_x, buf);
writeBinary(sum_y, buf);
writeBinary(sum_xx, buf);
writeBinary(sum_xy, buf);
}
void deserialize(ReadBuffer & buf)
{
readBinary(count, buf);
readBinary(sum_x, buf);
readBinary(sum_y, buf);
readBinary(sum_xx, buf);
readBinary(sum_xy, buf);
}
T getK() const
{
T divisor = sum_xx * count - sum_x * sum_x;
if (divisor == 0)
return std::numeric_limits<T>::quiet_NaN();
return (sum_xy * count - sum_x * sum_y) / divisor;
}
T getB(T k) const
{
if (count == 0)
return std::numeric_limits<T>::quiet_NaN();
return (sum_y - k * sum_x) / count;
}
};
/// Calculates simple linear regression parameters.
/// Result is a tuple (k, b) for y = k * x + b equation, solved by least squares approximation.
template <typename X, typename Y, typename Ret = Float64>
class AggregateFunctionSimpleLinearRegression final : public IAggregateFunctionDataHelper<
AggregateFunctionSimpleLinearRegressionData<Ret>,
AggregateFunctionSimpleLinearRegression<X, Y, Ret>
>
{
public:
AggregateFunctionSimpleLinearRegression(
const DataTypes & arguments,
const Array & params
):
IAggregateFunctionDataHelper<
AggregateFunctionSimpleLinearRegressionData<Ret>,
AggregateFunctionSimpleLinearRegression<X, Y, Ret>
> {arguments, params, createResultType()}
{
// notice: arguments has been checked before
}
String getName() const override
{
return "simpleLinearRegression";
}
void add(
AggregateDataPtr __restrict place,
const IColumn ** columns,
size_t row_num,
Arena *
) const override
{
auto col_x = assert_cast<const ColumnVector<X> *>(columns[0]);
auto col_y = assert_cast<const ColumnVector<Y> *>(columns[1]);
X x = col_x->getData()[row_num];
Y y = col_y->getData()[row_num];
this->data(place).add(x, y);
}
void merge(AggregateDataPtr __restrict place, ConstAggregateDataPtr rhs, Arena *) const override
{
this->data(place).merge(this->data(rhs));
}
void serialize(ConstAggregateDataPtr __restrict place, WriteBuffer & buf, std::optional<size_t> /* version */) const override
{
this->data(place).serialize(buf);
}
void deserialize(AggregateDataPtr __restrict place, ReadBuffer & buf, std::optional<size_t> /* version */, Arena *) const override
{
this->data(place).deserialize(buf);
}
static DataTypePtr createResultType()
{
DataTypes types
{
std::make_shared<DataTypeNumber<Ret>>(),
std::make_shared<DataTypeNumber<Ret>>(),
};
Strings names
{
"k",
"b",
};
return std::make_shared<DataTypeTuple>(
std::move(types),
std::move(names)
);
}
bool allocatesMemoryInArena() const override { return false; }
void insertResultInto(
AggregateDataPtr __restrict place,
IColumn & to,
Arena *) const override
{
Ret k = this->data(place).getK();
Ret b = this->data(place).getB(k);
auto & col_tuple = assert_cast<ColumnTuple &>(to);
auto & col_k = assert_cast<ColumnVector<Ret> &>(col_tuple.getColumn(0));
auto & col_b = assert_cast<ColumnVector<Ret> &>(col_tuple.getColumn(1));
col_k.getData().push_back(k);
col_b.getData().push_back(b);
}
};
}
|
