davidrichards-just_enumerable_stats 0.0.2
Sign up to get free protection for your applications and to get access to all the features.
- data/README.rdoc +147 -0
- data/VERSION.yml +4 -0
- data/bin/jes +27 -0
- data/lib/fixed_range.rb +46 -0
- data/lib/just_enumerable_stats/stats.rb +503 -0
- data/lib/just_enumerable_stats.rb +498 -0
- data/spec/fixed_range_spec.rb +77 -0
- data/spec/just_enumerable_stats/stats_spec.rb +459 -0
- data/spec/just_enumerable_stats_spec.rb +449 -0
- data/spec/spec_helper.rb +8 -0
- metadata +65 -0
@@ -0,0 +1,449 @@
|
|
1
|
+
require File.dirname(__FILE__) + '/spec_helper'
|
2
|
+
|
3
|
+
describe "JustEnumerableStats" do
|
4
|
+
|
5
|
+
before do
|
6
|
+
@a = [1,2,3]
|
7
|
+
@b = [8,4,2]
|
8
|
+
@doubler = lambda{|e| e * 2}
|
9
|
+
@inverser = lambda{|e| 1/e.to_f}
|
10
|
+
@inverse_matcher = lambda{|a, b| 1/a <=> 1/b}
|
11
|
+
end
|
12
|
+
|
13
|
+
it "should be able to generate a random number between two integers" do
|
14
|
+
val = (1..100).map {rand_between(1,10)}
|
15
|
+
(val.min >= 1).should be_true
|
16
|
+
(val.max <= 10).should be_true
|
17
|
+
end
|
18
|
+
|
19
|
+
it "should be able to generate a random number between two floats" do
|
20
|
+
val = (1..100).map {rand_in_floats(1.0,10.0)}
|
21
|
+
(val.min >= 1).should be_true
|
22
|
+
(val.max <= 10).should be_true
|
23
|
+
val.all? {|v| v.should be_is_a(Float)}
|
24
|
+
end
|
25
|
+
|
26
|
+
it "should be able to work with floats from rand_between" do
|
27
|
+
val = (1..100).map {rand_between(1.0,10.0)}
|
28
|
+
(val.min >= 1).should be_true
|
29
|
+
(val.max <= 10).should be_true
|
30
|
+
val.all? {|v| v.should be_is_a(Float)}
|
31
|
+
end
|
32
|
+
|
33
|
+
it "should have an original max" do
|
34
|
+
@a.original_max.should eql(3)
|
35
|
+
end
|
36
|
+
|
37
|
+
it "should have an original min" do
|
38
|
+
@a.original_min.should eql(1)
|
39
|
+
end
|
40
|
+
|
41
|
+
it "should have a max" do
|
42
|
+
@a.max.should eql(3)
|
43
|
+
end
|
44
|
+
|
45
|
+
it "should have a max that takes a block" do
|
46
|
+
val = @a.max &@inverse_matcher
|
47
|
+
val.should eql(1)
|
48
|
+
end
|
49
|
+
|
50
|
+
it "should be able to use a default block for max" do
|
51
|
+
@a.default_block = @inverse_matcher
|
52
|
+
@a.max.should eql(1)
|
53
|
+
end
|
54
|
+
|
55
|
+
it "should know the index of the max value" do
|
56
|
+
@a.max_index.should eql(2)
|
57
|
+
end
|
58
|
+
|
59
|
+
it "should find the first index value with max_index, in case there are duplicates" do
|
60
|
+
[1,2,3,3].max_index.should eql(2)
|
61
|
+
end
|
62
|
+
|
63
|
+
it "should use a block to find the max index" do
|
64
|
+
val = @a.max_index &@inverse_matcher
|
65
|
+
val.should eql(0)
|
66
|
+
end
|
67
|
+
|
68
|
+
it "should be able to use a default block to find the max index" do
|
69
|
+
@a.default_block = @inverse_matcher
|
70
|
+
@a.max_index.should eql(0)
|
71
|
+
end
|
72
|
+
|
73
|
+
it "should have a min" do
|
74
|
+
@a.min.should eql(1)
|
75
|
+
end
|
76
|
+
|
77
|
+
it "should have a min that takes a block" do
|
78
|
+
val = @a.min &@inverse_matcher
|
79
|
+
val.should eql(3)
|
80
|
+
end
|
81
|
+
|
82
|
+
it "should be able to use a default block for min" do
|
83
|
+
@a.default_block = @inverse_matcher
|
84
|
+
@a.min.should eql(3)
|
85
|
+
end
|
86
|
+
|
87
|
+
it "should know the index of the min value" do
|
88
|
+
@a.min_index.should eql(0)
|
89
|
+
end
|
90
|
+
|
91
|
+
it "should find the first index value with min_index, in case there are duplicates" do
|
92
|
+
[1,1,2,3].min_index.should eql(0)
|
93
|
+
end
|
94
|
+
|
95
|
+
it "should use a block to find the min index" do
|
96
|
+
val = @a.min_index &@inverse_matcher
|
97
|
+
val.should eql(2)
|
98
|
+
end
|
99
|
+
|
100
|
+
it "should be able to use a default block to find the min index" do
|
101
|
+
@a.default_block = @inverse_matcher
|
102
|
+
@a.min_index.should eql(2)
|
103
|
+
end
|
104
|
+
|
105
|
+
it "should be able to sum a list" do
|
106
|
+
@a.sum.should eql(6)
|
107
|
+
[1, 2, 3.0].sum.should eql(6.0)
|
108
|
+
end
|
109
|
+
|
110
|
+
it "should offer sum with a precision of 1.0e-15" do
|
111
|
+
[0.1, 0.2, 0.3].sum.should be_close(0.6, 1.0e-15)
|
112
|
+
[0.1, 0.2, 0.3].sum.should_not be_close(0.6, 1.0e-16)
|
113
|
+
end
|
114
|
+
|
115
|
+
it "should be able to evaluate a sum with a block" do
|
116
|
+
@a.sum(&@doubler).should eql(12)
|
117
|
+
end
|
118
|
+
|
119
|
+
it "should be able to use the default block to evaluate sum" do
|
120
|
+
@a.default_block = @doubler
|
121
|
+
@a.sum.should eql(12)
|
122
|
+
end
|
123
|
+
|
124
|
+
it "should be able to find the arithmetic average, mean, or avg" do
|
125
|
+
@a.average.should eql(2)
|
126
|
+
@a.mean.should eql(2)
|
127
|
+
@a.avg.should eql(2)
|
128
|
+
[1, 2, 3.0].average.should eql(2.0)
|
129
|
+
end
|
130
|
+
|
131
|
+
it "should be able to calculate average with a block" do
|
132
|
+
@a.average(&@doubler).should eql(4)
|
133
|
+
@a.mean(&@doubler).should eql(4)
|
134
|
+
@a.avg(&@doubler).should eql(4)
|
135
|
+
[1, 2, 3.0].average(&@doubler).should eql(4.0)
|
136
|
+
end
|
137
|
+
|
138
|
+
it "should be able to calculate average with a default block" do
|
139
|
+
@a.default_block = @doubler
|
140
|
+
@a.average.should eql(4)
|
141
|
+
@a.mean.should eql(4)
|
142
|
+
@a.avg.should eql(4)
|
143
|
+
b = [1, 2, 3.0]
|
144
|
+
b.default_block = @doubler
|
145
|
+
b.average.should eql(4.0)
|
146
|
+
end
|
147
|
+
|
148
|
+
it "should be able to calculate the variance" do
|
149
|
+
@a.variance.should eql(1)
|
150
|
+
@a.var.should eql(1)
|
151
|
+
end
|
152
|
+
|
153
|
+
it "should be able to calculate the variance with a block" do
|
154
|
+
@a.variance(&@doubler).should eql(4)
|
155
|
+
@a.var(&@doubler).should eql(4)
|
156
|
+
end
|
157
|
+
|
158
|
+
it "should be able to calculate the variance with a default block" do
|
159
|
+
@a.default_block = @doubler
|
160
|
+
@a.variance.should eql(4)
|
161
|
+
@a.var.should eql(4)
|
162
|
+
end
|
163
|
+
|
164
|
+
it "should be able to calculate the standard deviation" do
|
165
|
+
@a.standard_deviation.should eql(1.0)
|
166
|
+
@a.std.should eql(1.0)
|
167
|
+
end
|
168
|
+
|
169
|
+
it "should be able to calculate the standard deviation with a block" do
|
170
|
+
@a.standard_deviation(&@doubler).should eql(2.0)
|
171
|
+
@a.std(&@doubler).should eql(2.0)
|
172
|
+
end
|
173
|
+
|
174
|
+
it "should be able to calculate the standard deviation with a default block" do
|
175
|
+
@a.default_block = @doubler
|
176
|
+
@a.standard_deviation.should eql(2.0)
|
177
|
+
@a.std.should eql(2.0)
|
178
|
+
end
|
179
|
+
|
180
|
+
it "should be able to calculate the median value" do
|
181
|
+
@a.median.should eql(2)
|
182
|
+
[1,4,3,2,5].median.should eql(3)
|
183
|
+
[1,9,2,8].median.should eql(5.0)
|
184
|
+
end
|
185
|
+
|
186
|
+
it "should be able to get a max and min range from the list" do
|
187
|
+
@a.range.should eql([1, 3])
|
188
|
+
end
|
189
|
+
|
190
|
+
it "should be able to pass a block to the range method" do
|
191
|
+
@a.range(&@inverse_matcher).should eql([3, 1])
|
192
|
+
end
|
193
|
+
|
194
|
+
it "should be able to use a default block for the range method" do
|
195
|
+
@a.default_block = @inverse_matcher
|
196
|
+
@a.range.should eql([3, 1])
|
197
|
+
end
|
198
|
+
|
199
|
+
it "should have a getter and a setter on the range class" do
|
200
|
+
@a.range_class = Array
|
201
|
+
@a.range_class.should eql(Array)
|
202
|
+
end
|
203
|
+
|
204
|
+
it "should be able to instantiate a range" do
|
205
|
+
@a.range_as_range.should eql(Range.new(1, 3))
|
206
|
+
end
|
207
|
+
|
208
|
+
it "should be able to instantiate a range with a block" do
|
209
|
+
@a.range_as_range(&@inverse_matcher).should eql(Range.new(3, 1))
|
210
|
+
end
|
211
|
+
|
212
|
+
it "should be able to instantiate a range with a default block" do
|
213
|
+
@a.default_block = @inverse_matcher
|
214
|
+
@a.range_as_range.should eql(Range.new(3, 1))
|
215
|
+
end
|
216
|
+
|
217
|
+
it "should be able to create a new object, sorted" do
|
218
|
+
a = [3,1,2]
|
219
|
+
b = a.new_sort
|
220
|
+
b.object_id.should_not eql(a.object_id)
|
221
|
+
b.should eql([1,2,3])
|
222
|
+
a << 4
|
223
|
+
b.should eql([1,2,3])
|
224
|
+
end
|
225
|
+
|
226
|
+
it "should be able to take a block and still do new_sort" do
|
227
|
+
@a.new_sort(&@doubler).should eql([2,4,6])
|
228
|
+
end
|
229
|
+
|
230
|
+
it "should be able to take a default block and still do new_sort" do
|
231
|
+
@a.default_block = @doubler
|
232
|
+
@a.new_sort.should eql([2,4,6])
|
233
|
+
end
|
234
|
+
|
235
|
+
it "should be able to rank a list" do
|
236
|
+
@b.rank.should eql([3,2,1])
|
237
|
+
end
|
238
|
+
|
239
|
+
it "should be able to use a block in rank" do
|
240
|
+
@b.rank(&@inverser).should eql([1,2,3])
|
241
|
+
end
|
242
|
+
|
243
|
+
it "should be able to use a default block in rank" do
|
244
|
+
@b.default_block = @inverser
|
245
|
+
@b.rank.should eql([1,2,3])
|
246
|
+
end
|
247
|
+
|
248
|
+
it "should be able to get the order of values, handling duplicates" do
|
249
|
+
[10,5,5,1].order.should eql([4,2,3,1])
|
250
|
+
end
|
251
|
+
|
252
|
+
it "should be able to take a block in the ordering" do
|
253
|
+
[10,5,5,1].order(&@inverser).should eql([1,2,3,4])
|
254
|
+
end
|
255
|
+
|
256
|
+
it "should be able to take a default block in the ordering" do
|
257
|
+
a = [10,5,5,1]
|
258
|
+
a.default_block = @inverser
|
259
|
+
a.order.should eql([1,2,3,4])
|
260
|
+
end
|
261
|
+
|
262
|
+
it "should be able to calculate the cumulative sum" do
|
263
|
+
@a.cumulative_sum.should eql([1,3,6])
|
264
|
+
@a.cum_sum.should eql([1,3,6])
|
265
|
+
[1,2,3.0].cum_sum.should eql([1.0, 3.0, 6.0])
|
266
|
+
end
|
267
|
+
|
268
|
+
it "should be able to take a block to produce the cumulative sum" do
|
269
|
+
@a.cumulative_sum(&@doubler).should eql([2,6,12])
|
270
|
+
@a.cum_sum(&@doubler).should eql([2,6,12])
|
271
|
+
[1,2,3.0].cum_sum(&@doubler).should eql([2.0, 6.0, 12.0])
|
272
|
+
end
|
273
|
+
|
274
|
+
it "should be able to take a default block to produce the cumlative sum" do
|
275
|
+
@a.default_block = @doubler
|
276
|
+
@a.cumulative_sum.should eql([2,6,12])
|
277
|
+
@a.cum_sum.should eql([2,6,12])
|
278
|
+
b = [1,2,3.0]
|
279
|
+
b.default_block = @doubler
|
280
|
+
b.cum_sum.should eql([2.0, 6.0, 12.0])
|
281
|
+
end
|
282
|
+
|
283
|
+
it "should be able to calculate the cumulative product" do
|
284
|
+
@a.cumulative_product.should eql([1,2,6])
|
285
|
+
@a.cum_prod.should eql([1,2,6])
|
286
|
+
[1,2,3.0].cum_prod.should eql([1.0, 2.0, 6.0])
|
287
|
+
end
|
288
|
+
|
289
|
+
it "should be able to take a block to produce the cumulative product" do
|
290
|
+
@a.cumulative_product(&@doubler).should eql([2,8,48])
|
291
|
+
@a.cum_prod(&@doubler).should eql([2,8,48])
|
292
|
+
[1,2,3.0].cum_prod(&@doubler).should eql([2.0, 8.0, 48.0])
|
293
|
+
end
|
294
|
+
|
295
|
+
it "should be able to take a default block to produce the cumlative product" do
|
296
|
+
@a.default_block = @doubler
|
297
|
+
@a.cumulative_product.should eql([2,8,48])
|
298
|
+
@a.cum_prod.should eql([2,8,48])
|
299
|
+
b = [1,2,3.0]
|
300
|
+
b.default_block = @doubler
|
301
|
+
b.cum_prod.should eql([2.0, 8.0, 48.0])
|
302
|
+
end
|
303
|
+
|
304
|
+
it "should be able to produce the cumulative max" do
|
305
|
+
@a.cumulative_max.should eql([1,2,3])
|
306
|
+
@a.cum_max.should eql([1,2,3])
|
307
|
+
end
|
308
|
+
|
309
|
+
it "should be able to produce the cumulative max with a block" do
|
310
|
+
@a.cumulative_max(&@doubler).should eql([2,4,6])
|
311
|
+
@a.cum_max(&@doubler).should eql([2,4,6])
|
312
|
+
end
|
313
|
+
|
314
|
+
it "should be able to produce the cumulative max with a default block" do
|
315
|
+
@a.default_block = @doubler
|
316
|
+
@a.cumulative_max.should eql([2,4,6])
|
317
|
+
@a.cum_max.should eql([2,4,6])
|
318
|
+
end
|
319
|
+
|
320
|
+
it "should be able to produce the cumulative min" do
|
321
|
+
@a.cumulative_min.should eql([1,1,1])
|
322
|
+
@a.cum_min.should eql([1,1,1])
|
323
|
+
end
|
324
|
+
|
325
|
+
it "should be able to produce the cumulative min with a block" do
|
326
|
+
@a.cumulative_min(&@doubler).should eql([2,2,2])
|
327
|
+
@a.cum_min(&@doubler).should eql([2,2,2])
|
328
|
+
end
|
329
|
+
|
330
|
+
it "should be able to produce the cumulative min with a default block" do
|
331
|
+
@a.default_block = @doubler
|
332
|
+
@a.cumulative_min.should eql([2,2,2])
|
333
|
+
@a.cum_min.should eql([2,2,2])
|
334
|
+
end
|
335
|
+
|
336
|
+
it "should be able to multiply the values in the list" do
|
337
|
+
@a.product.should eql(6)
|
338
|
+
end
|
339
|
+
|
340
|
+
it "should be able to yield an operation on pairs" do
|
341
|
+
val = @a.to_pairs(@b) {|a, b| a + b}
|
342
|
+
val.should eql([9,6,5])
|
343
|
+
end
|
344
|
+
|
345
|
+
# [1,2,3] and [2,3,4] have 2 items in common, and 4 unique items together.
|
346
|
+
# So 2 items / 4 items is 0.5
|
347
|
+
it "should be able to find the tanimoto coefficient" do
|
348
|
+
b = [2,3,4]
|
349
|
+
@a.tanimoto_pairs(b).should eql(0.5)
|
350
|
+
@a.tanimoto_correlation(b).should eql(0.5)
|
351
|
+
end
|
352
|
+
|
353
|
+
it "should have long hand for union" do
|
354
|
+
@a.union(@b).should eql([1, 2, 3, 8, 4])
|
355
|
+
end
|
356
|
+
|
357
|
+
it "should have long hand for intersect" do
|
358
|
+
@a.intersect(@b).should eql([2])
|
359
|
+
end
|
360
|
+
|
361
|
+
it "should have a long hand for compliment" do
|
362
|
+
@a.compliment(@b).should eql([1, 3])
|
363
|
+
end
|
364
|
+
|
365
|
+
it "should have a long hand for exclusive not" do
|
366
|
+
@a.exclusive_not(@b).should eql([1,3,8,4])
|
367
|
+
end
|
368
|
+
|
369
|
+
it "should be able to generate a cartesian product" do
|
370
|
+
@a.cartesian_product(@b).should eql([[1, 8], [1, 4], [1, 2], [2, 8], [2, 4], [3, 8], [3, 4], [3, 2]])
|
371
|
+
@a.cp(@b).should eql([[1, 8], [1, 4], [1, 2], [2, 8], [2, 4], [3, 8], [3, 4], [3, 2]])
|
372
|
+
@a.permutations(@b).should eql([[1, 8], [1, 4], [1, 2], [2, 8], [2, 4], [3, 8], [3, 4], [3, 2]])
|
373
|
+
end
|
374
|
+
|
375
|
+
it "should be able to add pairwise computations" do
|
376
|
+
# Remember:
|
377
|
+
# @a = [1,2,3]
|
378
|
+
# @b = [8,4,2]
|
379
|
+
val = @a.sigma_pairs(@b) {|a, b| a / b}
|
380
|
+
val.should eql(1/8 + 2/4 + 3/2)
|
381
|
+
val = @a.sigma_pairs(@b) {|a, b| a * b}
|
382
|
+
val.should eql(1*8 + 2*4 + 3*2)
|
383
|
+
end
|
384
|
+
|
385
|
+
it "should be able to find the euclidian distance between two lists" do
|
386
|
+
@a.euclidian_distance(@b).should be_close(7.348, 0.001)
|
387
|
+
[1,2,3].euclidian_distance([2,3,4]).should eql(Math.sqrt(3.0))
|
388
|
+
end
|
389
|
+
|
390
|
+
it "should be able to generate a list of random numbers, each within the range between two lists" do
|
391
|
+
a = [1,0,-1]
|
392
|
+
b = [10,0,-20]
|
393
|
+
list_min = a.min_of_lists(b)
|
394
|
+
list_max = a.max_of_lists(b)
|
395
|
+
100.times do
|
396
|
+
val = a.rand_in_range(b)
|
397
|
+
val.each_with_index do |e, i|
|
398
|
+
e >= list_min[i]
|
399
|
+
e <= list_max[i]
|
400
|
+
end
|
401
|
+
end
|
402
|
+
end
|
403
|
+
|
404
|
+
it "should be able to find random numbers in the range between many lists" do
|
405
|
+
a = [1,0,-2]
|
406
|
+
b = [10,0,-20]
|
407
|
+
c = [2,0,-1]
|
408
|
+
list_min = a.min_of_lists(b)
|
409
|
+
list_max = a.max_of_lists(b)
|
410
|
+
100.times do
|
411
|
+
val = a.rand_in_range(b)
|
412
|
+
val.each_with_index do |e, i|
|
413
|
+
e >= list_min[i]
|
414
|
+
e <= list_max[i]
|
415
|
+
end
|
416
|
+
end
|
417
|
+
end
|
418
|
+
|
419
|
+
it "should be able to yield a block for columns of values" do
|
420
|
+
a = [1,2,6.0]
|
421
|
+
b = [2,6.0,1]
|
422
|
+
c = [6.0,1,2]
|
423
|
+
val = a.yield_transpose(b, c) {|e| e.mean}
|
424
|
+
val.should eql([3.0, 3.0, 3.0])
|
425
|
+
end
|
426
|
+
|
427
|
+
it "should be able to transpose a list of lists (not dependent on Array#transpose)" do
|
428
|
+
a = [1,2,6.0]
|
429
|
+
b = [2,6.0,1]
|
430
|
+
c = [6.0,1,2]
|
431
|
+
val = a.yield_transpose(b, c)
|
432
|
+
val.should eql( [[1, 2, 6.0], [2, 6.0, 1], [6.0, 1, 2]])
|
433
|
+
end
|
434
|
+
|
435
|
+
it "should be able to get the correlation between two lists" do
|
436
|
+
@a.correlation([2,3,5]).should be_close(0.981, 0.001)
|
437
|
+
@a.cor([2,3,5]).should be_close(0.981, 0.001)
|
438
|
+
end
|
439
|
+
|
440
|
+
it "should be able to return the max of lists" do
|
441
|
+
@a.max_of_lists(@b).should eql([8,4,3])
|
442
|
+
@a.max_of_lists(@b, [10,10,10]).should eql([10,10,10])
|
443
|
+
end
|
444
|
+
|
445
|
+
it "should be able to return the min of lists" do
|
446
|
+
@a.min_of_lists(@b).should eql([1,2,2])
|
447
|
+
end
|
448
|
+
|
449
|
+
end
|