line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
# |
2
|
|
|
|
|
|
|
# LSNoHistory.pm - least-squares regression without data history |
3
|
|
|
|
|
|
|
# |
4
|
|
|
|
|
|
|
# $Id: LSNoHistory.pm,v 1.6 2003/02/23 05:11:29 pliam Exp $ |
5
|
|
|
|
|
|
|
# |
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
package Statistics::LSNoHistory; |
8
|
3
|
|
|
3
|
|
72430
|
use strict; |
|
3
|
|
|
|
|
10
|
|
|
3
|
|
|
|
|
131
|
|
9
|
|
|
|
|
|
|
|
10
|
3
|
|
|
3
|
|
19
|
use vars qw($VERSION); |
|
3
|
|
|
|
|
5
|
|
|
3
|
|
|
|
|
5777
|
|
11
|
|
|
|
|
|
|
$VERSION = sprintf("%d.%02d", (q$Name: LSNoHist_Release_0_01 $ =~ /\d+/g)); |
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
############################################################################# |
14
|
|
|
|
|
|
|
# top-level pod |
15
|
|
|
|
|
|
|
############################################################################# |
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
=pod |
18
|
|
|
|
|
|
|
|
19
|
|
|
|
|
|
|
=head1 NAME |
20
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
Statistics::LSNoHistory - Least-Squares linear regression package without |
22
|
|
|
|
|
|
|
data history |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
=head1 SYNOPSIS |
25
|
|
|
|
|
|
|
|
26
|
|
|
|
|
|
|
# construct from points |
27
|
|
|
|
|
|
|
$reg = Statistics::LSNoHistory->new(points => [ |
28
|
|
|
|
|
|
|
1.0 => 1.0, |
29
|
|
|
|
|
|
|
2.1 => 1.9, |
30
|
|
|
|
|
|
|
2.8 => 3.2, |
31
|
|
|
|
|
|
|
4.0 => 4.1, |
32
|
|
|
|
|
|
|
5.2 => 4.9 |
33
|
|
|
|
|
|
|
]); |
34
|
|
|
|
|
|
|
|
35
|
|
|
|
|
|
|
# other equivalent constructions |
36
|
|
|
|
|
|
|
$reg = Statistics::LSNoHistory->new( |
37
|
|
|
|
|
|
|
xvalues => [1.0, 2.1, 2.8, 4.0, 5.2], |
38
|
|
|
|
|
|
|
yvalues => [1.0, 1.9, 3.2, 4.1, 4.9] |
39
|
|
|
|
|
|
|
); |
40
|
|
|
|
|
|
|
# or |
41
|
|
|
|
|
|
|
$reg = Statistics::LSNoHistory->new; |
42
|
|
|
|
|
|
|
$reg->append_arrays( |
43
|
|
|
|
|
|
|
[1.0, 2.1, 2.8, 4.0, 5.2], |
44
|
|
|
|
|
|
|
[1.0, 1.9, 3.2, 4.1, 4.9] |
45
|
|
|
|
|
|
|
); |
46
|
|
|
|
|
|
|
# or |
47
|
|
|
|
|
|
|
$reg = Statistics::LSNoHistory->new; |
48
|
|
|
|
|
|
|
$reg->append_points( |
49
|
|
|
|
|
|
|
1.0 => 1.0, 2.1 => 1.9, 2.8 => 3.2, 4.0 => 4.1, 5.2 => 4.9 |
50
|
|
|
|
|
|
|
); |
51
|
|
|
|
|
|
|
|
52
|
|
|
|
|
|
|
# You may also construct from the preliminary statistics of a |
53
|
|
|
|
|
|
|
# previous regression: |
54
|
|
|
|
|
|
|
$reg = Statistics::LSNoHistory->new( |
55
|
|
|
|
|
|
|
num => 5, |
56
|
|
|
|
|
|
|
sumx => 15.1, |
57
|
|
|
|
|
|
|
sumy => 15.1, |
58
|
|
|
|
|
|
|
sumxx => 56.29, |
59
|
|
|
|
|
|
|
sumyy => 55.67, |
60
|
|
|
|
|
|
|
sumxy => 55.83, |
61
|
|
|
|
|
|
|
minx => 1.0, |
62
|
|
|
|
|
|
|
maxx => 5.2, |
63
|
|
|
|
|
|
|
miny => 1.0, |
64
|
|
|
|
|
|
|
maxy => 4.9 |
65
|
|
|
|
|
|
|
); |
66
|
|
|
|
|
|
|
# thus a branch may be instantiated as follows |
67
|
|
|
|
|
|
|
$branch = Statistics::LSNoHistory->new(%{$reg->dump_stats}); |
68
|
|
|
|
|
|
|
$reg->append_point(6.1, 5.9); |
69
|
|
|
|
|
|
|
$branch->append_point(5.8, 6.0); |
70
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
# calculate regression values, print some |
72
|
|
|
|
|
|
|
printf("Slope: %.2f\n", $reg->slope); |
73
|
|
|
|
|
|
|
printf("Intercept %.2f\n", $reg->intercept); |
74
|
|
|
|
|
|
|
printf("Correlation Coefficient: %.2f\n", $reg->pearson_r); |
75
|
|
|
|
|
|
|
... |
76
|
|
|
|
|
|
|
|
77
|
|
|
|
|
|
|
|
78
|
|
|
|
|
|
|
=head1 DESCRIPTION |
79
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
This package provides standard least squares linear regression |
81
|
|
|
|
|
|
|
functionality without the need for storing the complete data history. |
82
|
|
|
|
|
|
|
Like any other, it finds best m,k (in least squares sense) so that |
83
|
|
|
|
|
|
|
y = m*x + k fits data points (x_1,y_1),...,(x_n,y_n). |
84
|
|
|
|
|
|
|
|
85
|
|
|
|
|
|
|
In many applications involving linear regression, it is desirable |
86
|
|
|
|
|
|
|
to compute a regression based on the intermediate statistics of a |
87
|
|
|
|
|
|
|
previous regression along with any I data points. Thus there |
88
|
|
|
|
|
|
|
is no need to store a complete data history, but rather only a minimal |
89
|
|
|
|
|
|
|
set of intermediate statistics, the number of which, thanks to Gauss, |
90
|
|
|
|
|
|
|
is 6. |
91
|
|
|
|
|
|
|
|
92
|
|
|
|
|
|
|
The user interface provides a way to instantiate a regression object |
93
|
|
|
|
|
|
|
with either raw data or previous intermediate statistics. |
94
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
=cut |
96
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
############################################################################# |
98
|
|
|
|
|
|
|
# construction |
99
|
|
|
|
|
|
|
############################################################################# |
100
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
=pod |
102
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
=head1 CONSTRUCTOR ARGUMENTS |
104
|
|
|
|
|
|
|
|
105
|
|
|
|
|
|
|
The constructor (or class method I) takes several possible |
106
|
|
|
|
|
|
|
arguments. The initialization scenario depends on the kinds of |
107
|
|
|
|
|
|
|
arguments passed and falls into one of the following categories: |
108
|
|
|
|
|
|
|
|
109
|
|
|
|
|
|
|
=over 2 |
110
|
|
|
|
|
|
|
|
111
|
|
|
|
|
|
|
=item * |
112
|
|
|
|
|
|
|
|
113
|
|
|
|
|
|
|
I S() by itself is equivalent to initializing with no |
114
|
|
|
|
|
|
|
data. All internal statistics are set to zero. |
115
|
|
|
|
|
|
|
|
116
|
|
|
|
|
|
|
=item * |
117
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
I new(I => [x_1 => y_1, x_2 => y_2,..., |
119
|
|
|
|
|
|
|
x_n => y_n]) processes the n specified data points. Note that |
120
|
|
|
|
|
|
|
points expects an array reference even though we've written it |
121
|
|
|
|
|
|
|
in "hash notation" for clarity. |
122
|
|
|
|
|
|
|
|
123
|
|
|
|
|
|
|
=item * |
124
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
I new(I => [x_1, x_2,..., x_n], |
126
|
|
|
|
|
|
|
I => [y_1, y_2,..., y_n]) is equivalent to the above. |
127
|
|
|
|
|
|
|
|
128
|
|
|
|
|
|
|
=item * |
129
|
|
|
|
|
|
|
|
130
|
|
|
|
|
|
|
I new(I) requires I of the |
131
|
|
|
|
|
|
|
following intermediate statistics: |
132
|
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
=over 6 |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
=item I |
136
|
|
|
|
|
|
|
|
137
|
|
|
|
|
|
|
S<=E> Number of points. |
138
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
=item I |
140
|
|
|
|
|
|
|
|
141
|
|
|
|
|
|
|
S<=E> Sum of x values. |
142
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
=item I |
144
|
|
|
|
|
|
|
|
145
|
|
|
|
|
|
|
S<=E> Sum of y values. |
146
|
|
|
|
|
|
|
|
147
|
|
|
|
|
|
|
=item I |
148
|
|
|
|
|
|
|
|
149
|
|
|
|
|
|
|
S<=E> Sum of x values squared. |
150
|
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
=item I |
152
|
|
|
|
|
|
|
|
153
|
|
|
|
|
|
|
S<=E> Sum of y values squared. |
154
|
|
|
|
|
|
|
|
155
|
|
|
|
|
|
|
=item I |
156
|
|
|
|
|
|
|
|
157
|
|
|
|
|
|
|
S<=E> Sum of x*y products. |
158
|
|
|
|
|
|
|
|
159
|
|
|
|
|
|
|
=item I |
160
|
|
|
|
|
|
|
|
161
|
|
|
|
|
|
|
S<=E> Minimum x value. |
162
|
|
|
|
|
|
|
|
163
|
|
|
|
|
|
|
=item I |
164
|
|
|
|
|
|
|
|
165
|
|
|
|
|
|
|
S<=E> Maximum x value. |
166
|
|
|
|
|
|
|
|
167
|
|
|
|
|
|
|
=item I |
168
|
|
|
|
|
|
|
|
169
|
|
|
|
|
|
|
S<=E> Minimum y value. |
170
|
|
|
|
|
|
|
|
171
|
|
|
|
|
|
|
=item I |
172
|
|
|
|
|
|
|
|
173
|
|
|
|
|
|
|
S<=E> Maximum y value. |
174
|
|
|
|
|
|
|
|
175
|
|
|
|
|
|
|
=back 6 |
176
|
|
|
|
|
|
|
|
177
|
|
|
|
|
|
|
=back 2 |
178
|
|
|
|
|
|
|
|
179
|
|
|
|
|
|
|
=cut |
180
|
|
|
|
|
|
|
|
181
|
|
|
|
|
|
|
## new constructor |
182
|
|
|
|
|
|
|
sub new { |
183
|
11
|
|
|
11
|
0
|
2447
|
my $class = shift; |
184
|
11
|
|
|
|
|
40
|
my %args = @_; |
185
|
11
|
|
|
|
|
17
|
my $self; |
186
|
11
|
|
|
|
|
35
|
my @stats = qw(num sumx sumy sumxx sumyy sumxy); |
187
|
11
|
|
|
|
|
28
|
push(@stats, qw(minx maxx miny maxy)); # min/max |
188
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
# if complete set of statistics, construct from previous state |
190
|
|
|
|
|
|
|
# if (@stats == scalar(grep {defined($args{$_})} @stats)) { |
191
|
11
|
100
|
|
|
|
23
|
if (@stats == grep {defined($args{$_})} @stats) { |
|
110
|
|
|
|
|
281
|
|
192
|
|
|
|
|
|
|
# reject unsupported arguments and combinations |
193
|
2
|
50
|
|
|
|
5
|
if (grep {defined($args{$_})} qw(points xvalues yvalues)) { |
|
6
|
|
|
|
|
15
|
|
194
|
0
|
|
|
|
|
0
|
die "Cannot give new data along with previous state."; |
195
|
|
|
|
|
|
|
} |
196
|
2
|
50
|
|
|
|
9
|
unless (@stats == keys %args) { |
197
|
0
|
|
|
|
|
0
|
die "Unknown constructor arguments."; |
198
|
|
|
|
|
|
|
} |
199
|
|
|
|
|
|
|
# check the number of points for consistency |
200
|
2
|
50
|
|
|
|
7
|
unless (abs(int($args{num})) == $args{num}) { |
201
|
0
|
|
|
|
|
0
|
die "Bad number of points: must be positive integer."; |
202
|
|
|
|
|
|
|
} |
203
|
2
|
|
|
|
|
3
|
$self = \%args; |
204
|
2
|
|
|
|
|
5
|
bless $self, $class; |
205
|
2
|
|
|
|
|
9
|
return $self; |
206
|
|
|
|
|
|
|
} |
207
|
|
|
|
|
|
|
# in any other case we're starting from scratch |
208
|
9
|
|
|
|
|
17
|
$self = {}; |
209
|
9
|
|
|
|
|
20
|
bless $self, $class; |
210
|
9
|
|
|
|
|
31
|
$self->_init; |
211
|
|
|
|
|
|
|
# x & y value array refs |
212
|
9
|
100
|
66
|
|
|
61
|
if (defined($args{xvalues}) && defined($args{yvalues})) { |
|
|
100
|
|
|
|
|
|
213
|
3
|
50
|
|
|
|
632
|
if (defined $args{points}) { |
214
|
0
|
|
|
|
|
0
|
die "Must give points or array values, but not both"; |
215
|
|
|
|
|
|
|
} |
216
|
3
|
50
|
|
|
|
10
|
unless (scalar(keys %args) == 2) { |
217
|
0
|
|
|
|
|
0
|
die "Unknown constructor arguments."; |
218
|
|
|
|
|
|
|
} |
219
|
3
|
|
|
|
|
12
|
$self->append_arrays($args{xvalues}, $args{yvalues}); |
220
|
|
|
|
|
|
|
} |
221
|
|
|
|
|
|
|
# (x,y) point array ref |
222
|
|
|
|
|
|
|
elsif (defined($args{points})) { |
223
|
3
|
50
|
|
|
|
7
|
if (grep {defined($args{$_})} qw(xvalues yvalues)) { |
|
6
|
|
|
|
|
29
|
|
224
|
0
|
|
|
|
|
0
|
die "Must give points or array values, but not both"; |
225
|
|
|
|
|
|
|
} |
226
|
3
|
50
|
|
|
|
15
|
unless (scalar(keys %args) == 1) { |
227
|
0
|
|
|
|
|
0
|
die "Unknown constructor arguments."; |
228
|
|
|
|
|
|
|
} |
229
|
3
|
|
|
|
|
6
|
$self->append_points(@{$args{points}}); |
|
3
|
|
|
|
|
16
|
|
230
|
|
|
|
|
|
|
} |
231
|
|
|
|
|
|
|
# default constructor (already initialized above) |
232
|
|
|
|
|
|
|
else { |
233
|
3
|
50
|
|
|
|
10
|
if (scalar(keys %args)) { |
234
|
0
|
|
|
|
|
0
|
die "Unknown constructor arguments."; |
235
|
|
|
|
|
|
|
} |
236
|
|
|
|
|
|
|
} |
237
|
9
|
|
|
|
|
41
|
return $self; |
238
|
|
|
|
|
|
|
} |
239
|
|
|
|
|
|
|
|
240
|
|
|
|
|
|
|
## _init in this context really means start with state of 0's |
241
|
|
|
|
|
|
|
sub _init { |
242
|
9
|
|
|
9
|
|
14
|
my $self = shift; |
243
|
9
|
|
|
|
|
26
|
my @stats = qw(num sumx sumy sumxx sumyy sumxy); |
244
|
9
|
|
|
|
|
23
|
push(@stats, qw(minx maxx miny maxy)); # min/max |
245
|
|
|
|
|
|
|
|
246
|
9
|
|
|
|
|
90
|
@$self{@stats} = (0) x scalar(@stats); |
247
|
|
|
|
|
|
|
} |
248
|
|
|
|
|
|
|
|
249
|
|
|
|
|
|
|
|
250
|
|
|
|
|
|
|
############################################################################# |
251
|
|
|
|
|
|
|
# other methods |
252
|
|
|
|
|
|
|
############################################################################# |
253
|
|
|
|
|
|
|
=pod |
254
|
|
|
|
|
|
|
|
255
|
|
|
|
|
|
|
=head1 METHODS |
256
|
|
|
|
|
|
|
|
257
|
|
|
|
|
|
|
=over 2 |
258
|
|
|
|
|
|
|
|
259
|
|
|
|
|
|
|
=cut |
260
|
|
|
|
|
|
|
|
261
|
|
|
|
|
|
|
# |
262
|
|
|
|
|
|
|
# adding data |
263
|
|
|
|
|
|
|
# |
264
|
|
|
|
|
|
|
|
265
|
|
|
|
|
|
|
## append_point |
266
|
|
|
|
|
|
|
=pod |
267
|
|
|
|
|
|
|
|
268
|
|
|
|
|
|
|
=item * |
269
|
|
|
|
|
|
|
|
270
|
|
|
|
|
|
|
I(x,y) process an additional data point. |
271
|
|
|
|
|
|
|
|
272
|
|
|
|
|
|
|
=cut |
273
|
|
|
|
|
|
|
sub append_point { |
274
|
64
|
|
|
64
|
0
|
1756
|
my $self = shift; |
275
|
64
|
|
|
|
|
71
|
my($x,$y) = @_; |
276
|
|
|
|
|
|
|
|
277
|
|
|
|
|
|
|
## will have to recompute regression |
278
|
64
|
|
|
|
|
84
|
$self->{cached} = 0; |
279
|
|
|
|
|
|
|
|
280
|
|
|
|
|
|
|
# min/max |
281
|
64
|
100
|
|
|
|
110
|
if ($self->{num}) { |
282
|
55
|
100
|
|
|
|
126
|
$self->{minx} = ($x < $self->{minx}) ? $x : $self->{minx}; |
283
|
55
|
100
|
|
|
|
120
|
$self->{maxx} = ($x > $self->{maxx}) ? $x : $self->{maxx}; |
284
|
55
|
100
|
|
|
|
112
|
$self->{miny} = ($y < $self->{miny}) ? $y : $self->{miny}; |
285
|
55
|
100
|
|
|
|
106
|
$self->{maxy} = ($y > $self->{maxy}) ? $y : $self->{maxy}; |
286
|
|
|
|
|
|
|
} |
287
|
|
|
|
|
|
|
else { |
288
|
9
|
|
|
|
|
17
|
$self->{minx} = $x; |
289
|
9
|
|
|
|
|
27
|
$self->{maxx} = $x; |
290
|
9
|
|
|
|
|
16
|
$self->{miny} = $y; |
291
|
9
|
|
|
|
|
12
|
$self->{maxy} = $y; |
292
|
|
|
|
|
|
|
} |
293
|
|
|
|
|
|
|
|
294
|
|
|
|
|
|
|
# classic stats |
295
|
64
|
|
|
|
|
71
|
$self->{num}++; |
296
|
64
|
|
|
|
|
80
|
$self->{sumx} += $x; |
297
|
64
|
|
|
|
|
69
|
$self->{sumy} += $y; |
298
|
64
|
|
|
|
|
102
|
$self->{sumxx} += $x**2; |
299
|
64
|
|
|
|
|
87
|
$self->{sumyy} += $y**2; |
300
|
64
|
|
|
|
|
140
|
$self->{sumxy} += $x*$y; |
301
|
|
|
|
|
|
|
} |
302
|
|
|
|
|
|
|
|
303
|
|
|
|
|
|
|
## append_points |
304
|
|
|
|
|
|
|
=pod |
305
|
|
|
|
|
|
|
|
306
|
|
|
|
|
|
|
=item * |
307
|
|
|
|
|
|
|
|
308
|
|
|
|
|
|
|
I(x_1 => y_1,..., x_n => y_n) process additional data points, |
309
|
|
|
|
|
|
|
which is equivalent to calling append_point() n times. |
310
|
|
|
|
|
|
|
|
311
|
|
|
|
|
|
|
=cut |
312
|
|
|
|
|
|
|
sub append_points { |
313
|
4
|
|
|
4
|
0
|
9
|
my $self = shift; |
314
|
4
|
|
|
|
|
23
|
my @points = @_; |
315
|
4
|
|
|
|
|
8
|
my $num = scalar(@points); |
316
|
|
|
|
|
|
|
|
317
|
4
|
50
|
|
|
|
20
|
if ($num % 2) { die "Incomplete list of points."; } |
|
0
|
|
|
|
|
0
|
|
318
|
|
|
|
|
|
|
|
319
|
4
|
|
|
|
|
9
|
$num /= 2; |
320
|
4
|
|
|
|
|
29
|
for (1..$num) { $self->append_point(splice(@points, 0, 2)); } |
|
30
|
|
|
|
|
103
|
|
321
|
|
|
|
|
|
|
} |
322
|
|
|
|
|
|
|
|
323
|
|
|
|
|
|
|
|
324
|
|
|
|
|
|
|
## append_arrays |
325
|
|
|
|
|
|
|
=pod |
326
|
|
|
|
|
|
|
|
327
|
|
|
|
|
|
|
=item * |
328
|
|
|
|
|
|
|
|
329
|
|
|
|
|
|
|
I([x_1, x_2,..., x_n], [y_1, y_2,..., y_n]) |
330
|
|
|
|
|
|
|
process additional data points given a pair x and y data array |
331
|
|
|
|
|
|
|
references. Also equivalent to calling append_point() n times. |
332
|
|
|
|
|
|
|
|
333
|
|
|
|
|
|
|
=cut |
334
|
|
|
|
|
|
|
sub append_arrays { |
335
|
4
|
|
|
4
|
0
|
11
|
my $self = shift; |
336
|
4
|
|
|
|
|
4
|
my ($xr, $yr) = @_; |
337
|
4
|
|
|
|
|
5
|
my ($xn, $yn); |
338
|
|
|
|
|
|
|
|
339
|
|
|
|
|
|
|
# check arg type |
340
|
4
|
50
|
33
|
|
|
22
|
unless ((ref($xr) eq 'ARRAY') && (ref($yr) eq 'ARRAY')) { |
341
|
0
|
|
|
|
|
0
|
die "Must pass pair of array references."; |
342
|
|
|
|
|
|
|
} |
343
|
|
|
|
|
|
|
|
344
|
|
|
|
|
|
|
# check that sizes match |
345
|
4
|
|
|
|
|
6
|
$xn = scalar(@$xr); |
346
|
4
|
|
|
|
|
4
|
$yn = scalar(@$yr); |
347
|
4
|
50
|
|
|
|
11
|
unless ($xn == $yn) { die "Incomplete list of points."; } |
|
0
|
|
|
|
|
0
|
|
348
|
|
|
|
|
|
|
|
349
|
4
|
|
|
|
|
8
|
for (1..$xn) { $self->append_point(shift(@$xr), shift(@$yr)); } |
|
30
|
|
|
|
|
61
|
|
350
|
|
|
|
|
|
|
} |
351
|
|
|
|
|
|
|
|
352
|
|
|
|
|
|
|
# |
353
|
|
|
|
|
|
|
# computing the regression |
354
|
|
|
|
|
|
|
# |
355
|
|
|
|
|
|
|
|
356
|
|
|
|
|
|
|
## _regress method -- done behind the scenes & considered private |
357
|
|
|
|
|
|
|
sub _regress { |
358
|
12
|
|
|
12
|
|
17
|
my $self = shift; |
359
|
12
|
|
|
|
|
18
|
my($n) = $self->{num}; |
360
|
12
|
|
|
|
|
67
|
my($dx) = $n*$self->{sumxx} - $self->{sumx}**2; |
361
|
12
|
|
|
|
|
638
|
my($dy) = $n*$self->{sumyy} - $self->{sumy}**2; |
362
|
|
|
|
|
|
|
|
363
|
|
|
|
|
|
|
# check that we have 2 points |
364
|
12
|
100
|
|
|
|
30
|
unless ($n >= 2) { die "Must have at least 2 points for regression."; } |
|
1
|
|
|
|
|
6
|
|
365
|
|
|
|
|
|
|
# check data for consistency |
366
|
11
|
50
|
33
|
|
|
62
|
unless (($dx!=0) && ($dy!=0)) { |
367
|
0
|
|
|
|
|
0
|
die "Inconsistent data: would divide by zero."; |
368
|
|
|
|
|
|
|
} |
369
|
|
|
|
|
|
|
|
370
|
|
|
|
|
|
|
# means and variances |
371
|
11
|
|
|
|
|
25
|
$self->{avgx} = $self->{sumx}/$n; |
372
|
11
|
|
|
|
|
21
|
$self->{avgy} = $self->{sumy}/$n; |
373
|
11
|
|
|
|
|
33
|
$self->{varx} = $dx/$n/($n-1); |
374
|
11
|
|
|
|
|
27
|
$self->{vary} = $dy/$n/($n-1); |
375
|
|
|
|
|
|
|
|
376
|
|
|
|
|
|
|
# slopes and intercepts |
377
|
11
|
|
|
|
|
40
|
$self->{mx} = ($n*$self->{sumxy} - $self->{sumx}*$self->{sumy})/$dx; |
378
|
11
|
|
|
|
|
32
|
$self->{kx} = $self->{avgy} - $self->{mx}*$self->{avgx}; |
379
|
11
|
|
|
|
|
30
|
$self->{my} = ($n*$self->{sumxy} - $self->{sumx}*$self->{sumy})/$dy; |
380
|
11
|
|
|
|
|
25
|
$self->{ky} = $self->{avgx} - $self->{my}*$self->{avgy}; |
381
|
|
|
|
|
|
|
|
382
|
|
|
|
|
|
|
# correlation coefficient (Pearson's r) and chi squared |
383
|
11
|
|
|
|
|
51
|
$self->{r} = ($n*$self->{sumxy} - $self->{sumx}*$self->{sumy}) |
384
|
|
|
|
|
|
|
/ sqrt($dx*$dy); |
385
|
11
|
|
|
|
|
49
|
$self->{chi2} = (1-$self->{r}**2)*$dy/$n; |
386
|
|
|
|
|
|
|
|
387
|
|
|
|
|
|
|
# flag that regression calculations are up to date |
388
|
11
|
|
|
|
|
51
|
$self->{cached} = 1; |
389
|
|
|
|
|
|
|
} |
390
|
|
|
|
|
|
|
|
391
|
|
|
|
|
|
|
# |
392
|
|
|
|
|
|
|
# presentation of stats, prediction |
393
|
|
|
|
|
|
|
# |
394
|
|
|
|
|
|
|
|
395
|
|
|
|
|
|
|
## average_x |
396
|
|
|
|
|
|
|
=pod |
397
|
|
|
|
|
|
|
|
398
|
|
|
|
|
|
|
=item * |
399
|
|
|
|
|
|
|
|
400
|
|
|
|
|
|
|
I returns the mean of the x values. |
401
|
|
|
|
|
|
|
|
402
|
|
|
|
|
|
|
=cut |
403
|
|
|
|
|
|
|
sub average_x { |
404
|
19
|
|
|
19
|
0
|
7041
|
my $self = shift; |
405
|
19
|
100
|
|
|
|
64
|
$self->_regress unless $self->{cached}; |
406
|
18
|
|
|
|
|
149
|
return $self->{avgx} |
407
|
|
|
|
|
|
|
} |
408
|
|
|
|
|
|
|
|
409
|
|
|
|
|
|
|
## average_y |
410
|
|
|
|
|
|
|
=pod |
411
|
|
|
|
|
|
|
|
412
|
|
|
|
|
|
|
=item * |
413
|
|
|
|
|
|
|
|
414
|
|
|
|
|
|
|
I returns the mean of the y values. |
415
|
|
|
|
|
|
|
|
416
|
|
|
|
|
|
|
=cut |
417
|
|
|
|
|
|
|
sub average_y { |
418
|
18
|
|
|
18
|
0
|
1163
|
my $self = shift; |
419
|
18
|
50
|
|
|
|
62
|
$self->_regress unless $self->{cached}; |
420
|
18
|
|
|
|
|
668
|
return $self->{avgy} |
421
|
|
|
|
|
|
|
} |
422
|
|
|
|
|
|
|
|
423
|
|
|
|
|
|
|
## variance_x |
424
|
|
|
|
|
|
|
=pod |
425
|
|
|
|
|
|
|
|
426
|
|
|
|
|
|
|
=item * |
427
|
|
|
|
|
|
|
|
428
|
|
|
|
|
|
|
I returns the (n-1)-style variance of the x values. |
429
|
|
|
|
|
|
|
|
430
|
|
|
|
|
|
|
=cut |
431
|
|
|
|
|
|
|
sub variance_x { |
432
|
18
|
|
|
18
|
0
|
1147
|
my $self = shift; |
433
|
18
|
50
|
|
|
|
53
|
$self->_regress unless $self->{cached}; |
434
|
18
|
|
|
|
|
123
|
return $self->{varx} |
435
|
|
|
|
|
|
|
} |
436
|
|
|
|
|
|
|
|
437
|
|
|
|
|
|
|
## variance_y |
438
|
|
|
|
|
|
|
=pod |
439
|
|
|
|
|
|
|
|
440
|
|
|
|
|
|
|
=item * |
441
|
|
|
|
|
|
|
|
442
|
|
|
|
|
|
|
I returns the (n-1)-style variance of the y values. |
443
|
|
|
|
|
|
|
|
444
|
|
|
|
|
|
|
=cut |
445
|
|
|
|
|
|
|
sub variance_y { |
446
|
18
|
|
|
18
|
0
|
1216
|
my $self = shift; |
447
|
18
|
50
|
|
|
|
52
|
$self->_regress unless $self->{cached}; |
448
|
18
|
|
|
|
|
155
|
return $self->{vary} |
449
|
|
|
|
|
|
|
} |
450
|
|
|
|
|
|
|
|
451
|
|
|
|
|
|
|
## slope |
452
|
|
|
|
|
|
|
=pod |
453
|
|
|
|
|
|
|
|
454
|
|
|
|
|
|
|
=item * |
455
|
|
|
|
|
|
|
|
456
|
|
|
|
|
|
|
I returns the slope m so that y = m*x + k is a least squares fit. |
457
|
|
|
|
|
|
|
Note that this is the least (y-y_avg)**2, and thus the standard slope. |
458
|
|
|
|
|
|
|
|
459
|
|
|
|
|
|
|
=cut |
460
|
|
|
|
|
|
|
sub slope { |
461
|
19
|
|
|
19
|
0
|
2283
|
my $self = shift; |
462
|
19
|
50
|
|
|
|
58
|
$self->_regress unless $self->{cached}; |
463
|
19
|
|
|
|
|
140
|
return $self->{mx} |
464
|
|
|
|
|
|
|
} |
465
|
|
|
|
|
|
|
|
466
|
|
|
|
|
|
|
## intercept |
467
|
|
|
|
|
|
|
=pod |
468
|
|
|
|
|
|
|
|
469
|
|
|
|
|
|
|
=item * |
470
|
|
|
|
|
|
|
|
471
|
|
|
|
|
|
|
I returns the intercept k so that y = m*x + k is a least squares |
472
|
|
|
|
|
|
|
fit. Note again that this is the least (y-y_avg)**2, and thus the |
473
|
|
|
|
|
|
|
standard intercept. |
474
|
|
|
|
|
|
|
|
475
|
|
|
|
|
|
|
=cut |
476
|
|
|
|
|
|
|
sub intercept { |
477
|
19
|
|
|
19
|
0
|
1172
|
my $self = shift; |
478
|
19
|
50
|
|
|
|
55
|
$self->_regress unless $self->{cached}; |
479
|
19
|
|
|
|
|
134
|
return $self->{kx} |
480
|
|
|
|
|
|
|
} |
481
|
|
|
|
|
|
|
|
482
|
|
|
|
|
|
|
## predict - predicte a y value given an x value |
483
|
|
|
|
|
|
|
=pod |
484
|
|
|
|
|
|
|
|
485
|
|
|
|
|
|
|
=item * |
486
|
|
|
|
|
|
|
|
487
|
|
|
|
|
|
|
I(x) predicts a y value, given an x value. Computes m*x + k, where |
488
|
|
|
|
|
|
|
m, k are the standard regression slope and intercept (->slope and ->intercept, |
489
|
|
|
|
|
|
|
respectively) for the most recent data. |
490
|
|
|
|
|
|
|
|
491
|
|
|
|
|
|
|
=cut |
492
|
|
|
|
|
|
|
sub predict { |
493
|
0
|
|
|
0
|
0
|
0
|
my $self = shift; |
494
|
0
|
|
|
|
|
0
|
my($x) = @_; |
495
|
|
|
|
|
|
|
|
496
|
0
|
0
|
|
|
|
0
|
$self->_regress unless $self->{cached}; |
497
|
0
|
|
|
|
|
0
|
return $self->{mx}*$x + $self->{kx}; |
498
|
|
|
|
|
|
|
} |
499
|
|
|
|
|
|
|
|
500
|
|
|
|
|
|
|
## slope_y |
501
|
|
|
|
|
|
|
=pod |
502
|
|
|
|
|
|
|
|
503
|
|
|
|
|
|
|
=item * |
504
|
|
|
|
|
|
|
|
505
|
|
|
|
|
|
|
I returns the slope m' so that y = m'*x + k' is a least squares fit. |
506
|
|
|
|
|
|
|
Note that this is the least (x-x_avg)**2, and thus I the standard slope. |
507
|
|
|
|
|
|
|
|
508
|
|
|
|
|
|
|
=cut |
509
|
|
|
|
|
|
|
sub slope_y { |
510
|
18
|
|
|
18
|
0
|
1182
|
my $self = shift; |
511
|
18
|
50
|
|
|
|
49
|
$self->_regress unless $self->{cached}; |
512
|
18
|
|
|
|
|
129
|
return $self->{my} |
513
|
|
|
|
|
|
|
} |
514
|
|
|
|
|
|
|
|
515
|
|
|
|
|
|
|
## intercept_y |
516
|
|
|
|
|
|
|
=pod |
517
|
|
|
|
|
|
|
|
518
|
|
|
|
|
|
|
=item * |
519
|
|
|
|
|
|
|
|
520
|
|
|
|
|
|
|
I returns the intercept k' so that y = m'*x + k' is a least |
521
|
|
|
|
|
|
|
squares fit. Note that this is the least (x-x_avg)**2, and thus I |
522
|
|
|
|
|
|
|
the standard intercept. |
523
|
|
|
|
|
|
|
|
524
|
|
|
|
|
|
|
=cut |
525
|
|
|
|
|
|
|
sub intercept_y { |
526
|
18
|
|
|
18
|
0
|
1120
|
my $self = shift; |
527
|
18
|
50
|
|
|
|
46
|
$self->_regress unless $self->{cached}; |
528
|
18
|
|
|
|
|
134
|
return $self->{ky} |
529
|
|
|
|
|
|
|
} |
530
|
|
|
|
|
|
|
|
531
|
|
|
|
|
|
|
## predict_x - predicte an x value given a y value |
532
|
|
|
|
|
|
|
=pod |
533
|
|
|
|
|
|
|
|
534
|
|
|
|
|
|
|
=item * |
535
|
|
|
|
|
|
|
|
536
|
|
|
|
|
|
|
I(y) predicts an x value given a y value. Computes m'*y + k', |
537
|
|
|
|
|
|
|
where m', k' are the regression (y-reletive) slope and intercept |
538
|
|
|
|
|
|
|
(->slope_y and ->intercept_y, respectively) for the most recent data. |
539
|
|
|
|
|
|
|
|
540
|
|
|
|
|
|
|
=cut |
541
|
|
|
|
|
|
|
sub predict_x { |
542
|
0
|
|
|
0
|
0
|
0
|
my $self = shift; |
543
|
0
|
|
|
|
|
0
|
my($y) = @_; |
544
|
|
|
|
|
|
|
|
545
|
0
|
0
|
|
|
|
0
|
$self->_regress unless $self->{cached}; |
546
|
0
|
|
|
|
|
0
|
return $self->{my}*$y + $self->{ky}; |
547
|
|
|
|
|
|
|
} |
548
|
|
|
|
|
|
|
|
549
|
|
|
|
|
|
|
## pearson_r |
550
|
|
|
|
|
|
|
=pod |
551
|
|
|
|
|
|
|
|
552
|
|
|
|
|
|
|
=item * |
553
|
|
|
|
|
|
|
|
554
|
|
|
|
|
|
|
I returns Pearson's r correlation coefficient. |
555
|
|
|
|
|
|
|
|
556
|
|
|
|
|
|
|
=cut |
557
|
|
|
|
|
|
|
sub pearson_r { |
558
|
21
|
|
|
21
|
0
|
1925
|
my $self = shift; |
559
|
21
|
100
|
|
|
|
59
|
$self->_regress unless $self->{cached}; |
560
|
21
|
|
|
|
|
158
|
return $self->{r} |
561
|
|
|
|
|
|
|
} |
562
|
|
|
|
|
|
|
|
563
|
|
|
|
|
|
|
## chi_squared |
564
|
|
|
|
|
|
|
=pod |
565
|
|
|
|
|
|
|
|
566
|
|
|
|
|
|
|
=item * |
567
|
|
|
|
|
|
|
|
568
|
|
|
|
|
|
|
I returns the chi squared statistic. |
569
|
|
|
|
|
|
|
|
570
|
|
|
|
|
|
|
=cut |
571
|
|
|
|
|
|
|
sub chi_squared { |
572
|
17
|
|
|
17
|
0
|
1187
|
my $self = shift; |
573
|
17
|
50
|
|
|
|
47
|
$self->_regress unless $self->{cached}; |
574
|
17
|
|
|
|
|
140
|
return $self->{chi2} |
575
|
|
|
|
|
|
|
} |
576
|
|
|
|
|
|
|
|
577
|
|
|
|
|
|
|
## minimum_x |
578
|
|
|
|
|
|
|
=pod |
579
|
|
|
|
|
|
|
|
580
|
|
|
|
|
|
|
=item * |
581
|
|
|
|
|
|
|
|
582
|
|
|
|
|
|
|
I returns the minimum x value |
583
|
|
|
|
|
|
|
|
584
|
|
|
|
|
|
|
=cut |
585
|
17
|
|
|
17
|
0
|
1768
|
sub minimum_x { return shift->{minx}; } |
586
|
|
|
|
|
|
|
|
587
|
|
|
|
|
|
|
## maximum_x |
588
|
|
|
|
|
|
|
=pod |
589
|
|
|
|
|
|
|
|
590
|
|
|
|
|
|
|
=item * |
591
|
|
|
|
|
|
|
|
592
|
|
|
|
|
|
|
I returns the maximum x value |
593
|
|
|
|
|
|
|
|
594
|
|
|
|
|
|
|
=cut |
595
|
17
|
|
|
17
|
0
|
1512
|
sub maximum_x { return shift->{maxx}; } |
596
|
|
|
|
|
|
|
|
597
|
|
|
|
|
|
|
## minimum_y |
598
|
|
|
|
|
|
|
=pod |
599
|
|
|
|
|
|
|
|
600
|
|
|
|
|
|
|
=item * |
601
|
|
|
|
|
|
|
|
602
|
|
|
|
|
|
|
I returns the minimum y value |
603
|
|
|
|
|
|
|
|
604
|
|
|
|
|
|
|
=cut |
605
|
17
|
|
|
17
|
0
|
1527
|
sub minimum_y { return shift->{miny}; } |
606
|
|
|
|
|
|
|
|
607
|
|
|
|
|
|
|
## maximum_y |
608
|
|
|
|
|
|
|
=pod |
609
|
|
|
|
|
|
|
|
610
|
|
|
|
|
|
|
=item * |
611
|
|
|
|
|
|
|
|
612
|
|
|
|
|
|
|
I returns the maximum y value |
613
|
|
|
|
|
|
|
|
614
|
|
|
|
|
|
|
=cut |
615
|
17
|
|
|
17
|
0
|
1405
|
sub maximum_y { return shift->{maxy}; } |
616
|
|
|
|
|
|
|
|
617
|
|
|
|
|
|
|
## dump_stats |
618
|
|
|
|
|
|
|
=pod |
619
|
|
|
|
|
|
|
|
620
|
|
|
|
|
|
|
=item * |
621
|
|
|
|
|
|
|
|
622
|
|
|
|
|
|
|
I returns a hash reference of the form |
623
|
|
|
|
|
|
|
|
624
|
|
|
|
|
|
|
{ num => , |
625
|
|
|
|
|
|
|
sumx => , |
626
|
|
|
|
|
|
|
sumy => , |
627
|
|
|
|
|
|
|
sumxx => , |
628
|
|
|
|
|
|
|
sumyy => , |
629
|
|
|
|
|
|
|
sumxy => , |
630
|
|
|
|
|
|
|
minx => , |
631
|
|
|
|
|
|
|
maxx => , |
632
|
|
|
|
|
|
|
miny => , |
633
|
|
|
|
|
|
|
maxy => } |
634
|
|
|
|
|
|
|
|
635
|
|
|
|
|
|
|
in other words, containing all the stats required by the final constructor |
636
|
|
|
|
|
|
|
above. This effectively dumps the regression history. |
637
|
|
|
|
|
|
|
|
638
|
|
|
|
|
|
|
=cut |
639
|
|
|
|
|
|
|
sub dump_stats { |
640
|
11
|
|
|
11
|
0
|
1792
|
my $self = shift; |
641
|
11
|
|
|
|
|
31
|
my @stats = qw(num sumx sumy sumxx sumyy sumxy); |
642
|
11
|
|
|
|
|
22
|
push(@stats, qw(minx maxx miny maxy)); # min/max |
643
|
11
|
|
|
|
|
21
|
my %dump; |
644
|
|
|
|
|
|
|
|
645
|
11
|
|
|
|
|
119
|
@dump{@stats} = @$self{@stats}; |
646
|
11
|
|
|
|
|
63
|
return \%dump; |
647
|
|
|
|
|
|
|
} |
648
|
|
|
|
|
|
|
|
649
|
|
|
|
|
|
|
1; |
650
|
|
|
|
|
|
|
|
651
|
|
|
|
|
|
|
__END__ |