1 | [[tags: egg]] |
---|
2 | [[toc:]] |
---|
3 | |
---|
4 | == Statistics |
---|
5 | |
---|
6 | This library is a port of [[http://compbio.ucdenver.edu/hunter/|Larry Hunter]]'s Lisp statistics library to chicken scheme. |
---|
7 | |
---|
8 | The library provides a number of formulae and methods taken from the book "Fundamentals of Biostatistics" by Bernard Rosner (5th edition). |
---|
9 | |
---|
10 | === Statistical Distributions |
---|
11 | |
---|
12 | To use this library, you need to understand the underlying statistics. In brief: |
---|
13 | |
---|
14 | The [[http://en.wikipedia.org/wiki/Binomial_distribution|Binomial |
---|
15 | distribution]] is used when counting discrete events in a series of |
---|
16 | trials, each of which events has a probability p of producing a |
---|
17 | positive outcome. An example would be tossing a coin {{n}} times: the |
---|
18 | probability of a head is {{p}}, and the distribution gives the |
---|
19 | expected number of heads in the {{n}} trials. The binomial |
---|
20 | distribution is defined as B(n, p). |
---|
21 | |
---|
22 | The [[http://en.wikipedia.org/wiki/Poisson_distribution|Poisson |
---|
23 | distribution]] is used to count discrete events which occur with a |
---|
24 | known average rate. A typical example is the decay of radioactive |
---|
25 | elements. A poisson distribution is defined Pois(mu). |
---|
26 | |
---|
27 | The [[http://en.wikipedia.org/wiki/Normal_distribution|Normal |
---|
28 | distribution]] is used for real-valued events which cluster around a |
---|
29 | specific mean with a symmetric variance. A typical example would be |
---|
30 | the distribution of people's heights. A normal distribution is |
---|
31 | defined N(mean, variance). |
---|
32 | |
---|
33 | === Provided Functions |
---|
34 | |
---|
35 | ==== Utilities |
---|
36 | |
---|
37 | <procedure>(average-rank value sorted-values)</procedure> |
---|
38 | returns the average position of given value in the list of sorted values: the rank is based from 1. |
---|
39 | > (average-rank 2 '(1 2 2 3 4)) |
---|
40 | 5/2 |
---|
41 | |
---|
42 | <procedure>(beta-incomplete x a b)</procedure> |
---|
43 | |
---|
44 | <procedure>(bin-and-count items n)</procedure> |
---|
45 | Divides the range of the list of {{items}} into {{n}} bins, and returns a vector of the number of items which fall into each bin. |
---|
46 | > (bin-and-count '(1 1 2 3 3 4 5) 5) |
---|
47 | #(2 1 2 1 1) |
---|
48 | |
---|
49 | <procedure>(combinations n k)</procedure> |
---|
50 | returns the number of ways to select {{k}} items from {{n}}, where the order does not matter. |
---|
51 | |
---|
52 | <procedure>(factorial n)</procedure> |
---|
53 | returns the factorial of {{n}}. |
---|
54 | |
---|
55 | <procedure>(find-critical-value p-function p-value #:increasing?)</procedure> |
---|
56 | given a monotonic function {{p-function}} taking a single value {{x}} to {{y}}, returns the value of {{x}} which makes {{(p-function x)}} closest to {{p-value}}. A boolean keyword parameter {{#:increasing?}} determines if function should be increasing or decreasing (the default). |
---|
57 | |
---|
58 | <procedure>(fisher-z-transform r)</procedure> |
---|
59 | returns the transformation of a correlation coefficient {{r}} into an approximately normal distribution. |
---|
60 | |
---|
61 | <procedure>(gamma-incomplete a x)</procedure> |
---|
62 | |
---|
63 | <procedure>(gamma-ln x)</procedure> |
---|
64 | |
---|
65 | <procedure>(permutations n k)</procedure> |
---|
66 | returns the number of ways to select {{k}} items from {{n}}, where the order does matter. |
---|
67 | |
---|
68 | <procedure>(random-normal mean sd)</procedure> |
---|
69 | returns a random number distributed with specified mean and standard deviation. |
---|
70 | |
---|
71 | <procedure>(random-pick items)</procedure> |
---|
72 | returns a random item from the given list of items. |
---|
73 | |
---|
74 | <procedure>(random-sample n items)</procedure> |
---|
75 | returns a random sample from the list of items without replacement of size {{n}}. |
---|
76 | |
---|
77 | <procedure>(random-weighted-sample n items weights)</procedure> |
---|
78 | returns a random sample from the list of items without replacement of size {{n}}, where each sample has a defined probability of selection (weight). |
---|
79 | |
---|
80 | <procedure>(sign n)</procedure> |
---|
81 | returns 0, 1 or -1 according to if {{n}} is zero, positive or negative. |
---|
82 | |
---|
83 | <procedure>(square n)</procedure> |
---|
84 | |
---|
85 | <procedure>(cumsum sequences)</procedure> |
---|
86 | returns the cumulative sum of a sequence. |
---|
87 | |
---|
88 | ==== Descriptive statistics |
---|
89 | |
---|
90 | These functions provide information on a given list of numbers, the {{items}}. Note, the list does not have to be sorted. |
---|
91 | |
---|
92 | <procedure>(mean items)</procedure> |
---|
93 | returns the arithmetic mean of the {{items}} (the sum of the numbers divided by the number of numbers). |
---|
94 | (mean '(1 2 3 4 5)) => 3 |
---|
95 | |
---|
96 | <procedure>(median items)</procedure> |
---|
97 | returns the value which separates the upper and lower halves of the list of numbers. |
---|
98 | (median '(1 2 3 4)) => 5/2 |
---|
99 | |
---|
100 | <procedure>(mode items)</procedure> |
---|
101 | returns two '''values'''. The first is a list of the ''modes'' and the second is the frequency. (A mode of a list of numbers is the most frequently occurring value.) |
---|
102 | > (mode '(1 2 3 4)) |
---|
103 | (1 2 3 4) |
---|
104 | 1 |
---|
105 | > (mode '(1 2 2 3 4)) |
---|
106 | (2) |
---|
107 | 2 |
---|
108 | > (mode '(1 2 2 3 3 4)) |
---|
109 | (2 3) |
---|
110 | 2 |
---|
111 | |
---|
112 | <procedure>(geometric-mean items)</procedure> |
---|
113 | returns the geometric mean of the {{items}} (the result of multiplying the items together and then taking the nth root, where n is the number of items). |
---|
114 | (geometric-mean '(1 2 3 4 5)) => 2.60517108469735 |
---|
115 | |
---|
116 | <procedure>(range items)</procedure> |
---|
117 | returns the difference between the biggest and the smallest value from the list of {{items}}. |
---|
118 | (range '(5 1 2 3 4)) => 4 |
---|
119 | |
---|
120 | <procedure>(percentile items percent)</procedure> |
---|
121 | returns the item closest to the {{percent}} value if the {{items}} are sorted into order; the returned item may be in the list, or the average of adjacent values. |
---|
122 | (percentile '(1 2 3 4) 50) => 5/2 |
---|
123 | (percentile '(1 2 3 4) 67) => 3 |
---|
124 | |
---|
125 | <procedure>(variance items)</procedure> |
---|
126 | |
---|
127 | <procedure>(standard-deviation items)</procedure> |
---|
128 | |
---|
129 | <procedure>(coefficient-of-variation items)</procedure> |
---|
130 | returns 100 * (std-dev / mean) of the {{items}}. |
---|
131 | (coefficient-of-variation '(1 2 3 4)) => 51.6397779494322 |
---|
132 | |
---|
133 | <procedure>(standard-error-of-the-mean items)</procedure> |
---|
134 | returns std-dev / sqrt(length items). |
---|
135 | (standard-error-of-the-mean '(1 2 3 4)) => 0.645497224367903 |
---|
136 | |
---|
137 | <procedure>(mean-sd-n items)</procedure> |
---|
138 | returns three '''values''', one for the mean, one for the standard deviation, and one for the length of the list. |
---|
139 | > (mean-sd-n '(1 2 3 4)) |
---|
140 | 5/2 |
---|
141 | 1.29099444873581 |
---|
142 | 4 |
---|
143 | |
---|
144 | ==== Distributional functions |
---|
145 | |
---|
146 | <procedure>(binomial-probability n k p)</procedure> |
---|
147 | returns the probability that the number of positive outcomes for a binomial distribution B(n, p) is k. |
---|
148 | > (do-ec (: i 0 11) |
---|
149 | (format #t "i = ~d P = ~f~&" i (binomial-probability 10 i 0.5))) |
---|
150 | i = 0 P = 0.0009765625 |
---|
151 | i = 1 P = 0.009765625 |
---|
152 | i = 2 P = 0.0439453125 |
---|
153 | i = 3 P = 0.1171875 |
---|
154 | i = 4 P = 0.205078125 |
---|
155 | i = 5 P = 0.24609375 |
---|
156 | i = 6 P = 0.205078125 |
---|
157 | i = 7 P = 0.1171875 |
---|
158 | i = 8 P = 0.0439453125 |
---|
159 | i = 9 P = 0.009765625 |
---|
160 | i = 10 P = 0.0009765625 |
---|
161 | |
---|
162 | <procedure>(binomial-cumulative-probability n k p)</procedure> |
---|
163 | returns the probability that less than {{k}} positive outcomes occur for a binomial distribution B(n, p). |
---|
164 | > (do-ec (: i 0 11) |
---|
165 | (format #t "i = ~d P = ~f~&" i (binomial-cumulative-probability 10 i 0.5))) |
---|
166 | i = 0 P = 0.0 |
---|
167 | i = 1 P = 0.0009765625 |
---|
168 | i = 2 P = 0.0107421875 |
---|
169 | i = 3 P = 0.0546875 |
---|
170 | i = 4 P = 0.171875 |
---|
171 | i = 5 P = 0.376953125 |
---|
172 | i = 6 P = 0.623046875 |
---|
173 | i = 7 P = 0.828125 |
---|
174 | i = 8 P = 0.9453125 |
---|
175 | i = 9 P = 0.9892578125 |
---|
176 | i = 10 P = 0.9990234375 |
---|
177 | |
---|
178 | <procedure>(binomial-ge-probability n k p)</procedure> |
---|
179 | returns the probability of {{k}} or more positive outcomes for a binomial distribution B(n, p). |
---|
180 | |
---|
181 | <procedure>(binomial-le-probability n k p)</procedure> |
---|
182 | returns the probability {{k}} or fewer positive outcomes for a binomial distribution B(n, p). |
---|
183 | |
---|
184 | <procedure>(poisson-probability mu k)</procedure> |
---|
185 | returns the probability of {{k}} events occurring when the average is {{mu}}. |
---|
186 | > (do-ec (: i 0 20) |
---|
187 | (format #t "P(X=~2d) = ~,4f~&" i (poisson-probability 10 i))) |
---|
188 | P(X= 0) = 0.0000 |
---|
189 | P(X= 1) = 0.0005 |
---|
190 | P(X= 2) = 0.0023 |
---|
191 | P(X= 3) = 0.0076 |
---|
192 | P(X= 4) = 0.0189 |
---|
193 | P(X= 5) = 0.0378 |
---|
194 | P(X= 6) = 0.0631 |
---|
195 | P(X= 7) = 0.0901 |
---|
196 | P(X= 8) = 0.1126 |
---|
197 | P(X= 9) = 0.1251 |
---|
198 | P(X=10) = 0.1251 |
---|
199 | P(X=11) = 0.1137 |
---|
200 | P(X=12) = 0.0948 |
---|
201 | P(X=13) = 0.0729 |
---|
202 | P(X=14) = 0.0521 |
---|
203 | P(X=15) = 0.0347 |
---|
204 | P(X=16) = 0.0217 |
---|
205 | P(X=17) = 0.0128 |
---|
206 | P(X=18) = 0.0071 |
---|
207 | P(X=19) = 0.0037 |
---|
208 | |
---|
209 | <procedure>(poisson-cumulative-probability mu k)</procedure> |
---|
210 | returns the probability of less than {{k}} events occurring when the average is {{mu}}. |
---|
211 | > (do-ec (: i 0 20) |
---|
212 | (format #t "P(X=~2d) = ~,4f~&" i (poisson-cumulative-probability 10 i))) |
---|
213 | P(X= 0) = 0.0000 |
---|
214 | P(X= 1) = 0.0000 |
---|
215 | P(X= 2) = 0.0005 |
---|
216 | P(X= 3) = 0.0028 |
---|
217 | P(X= 4) = 0.0103 |
---|
218 | P(X= 5) = 0.0293 |
---|
219 | P(X= 6) = 0.0671 |
---|
220 | P(X= 7) = 0.1301 |
---|
221 | P(X= 8) = 0.2202 |
---|
222 | P(X= 9) = 0.3328 |
---|
223 | P(X=10) = 0.4579 |
---|
224 | P(X=11) = 0.5830 |
---|
225 | P(X=12) = 0.6968 |
---|
226 | P(X=13) = 0.7916 |
---|
227 | P(X=14) = 0.8645 |
---|
228 | P(X=15) = 0.9165 |
---|
229 | P(X=16) = 0.9513 |
---|
230 | P(X=17) = 0.9730 |
---|
231 | P(X=18) = 0.9857 |
---|
232 | P(X=19) = 0.9928 |
---|
233 | |
---|
234 | <procedure>(poisson-ge-probability mu k)</procedure> |
---|
235 | returns the probability of {{k}} or more events occurring when the average is {{mu}}. |
---|
236 | |
---|
237 | <procedure>(normal-pdf x mean variance)</procedure> |
---|
238 | returns the likelihood of {{x}} given a normal distribution with stated mean and variance. |
---|
239 | > (do-ec (: i 0 11) |
---|
240 | (format #t "~3d ~,4f~&" i (normal-pdf i 5 4))) |
---|
241 | 0 0.0088 |
---|
242 | 1 0.0270 |
---|
243 | 2 0.0648 |
---|
244 | 3 0.1210 |
---|
245 | 4 0.1760 |
---|
246 | 5 0.1995 |
---|
247 | 6 0.1760 |
---|
248 | 7 0.1210 |
---|
249 | 8 0.0648 |
---|
250 | 9 0.0270 |
---|
251 | 10 0.0088 |
---|
252 | |
---|
253 | <procedure>(convert-to-standard-normal x mean variance)</procedure> |
---|
254 | returns a value for {{x}} rescaling the given normal distribution to a standard N(0, 1). |
---|
255 | > (convert-to-standard-normal 5 6 2) |
---|
256 | -1/2 |
---|
257 | |
---|
258 | <procedure>(phi x)</procedure> |
---|
259 | returns the cumulative distribution function (CDF) of the standard normal distribution. |
---|
260 | > (do-ec (: x -2 2 0.4) |
---|
261 | (format #t "~4,1f ~,4f~&" x (phi x))) |
---|
262 | -2.0 0.0228 |
---|
263 | -1.6 0.0548 |
---|
264 | -1.2 0.1151 |
---|
265 | -0.8 0.2119 |
---|
266 | -0.4 0.3446 |
---|
267 | 0.0 0.5000 |
---|
268 | 0.4 0.6554 |
---|
269 | 0.8 0.7881 |
---|
270 | 1.2 0.8849 |
---|
271 | 1.6 0.9452 |
---|
272 | |
---|
273 | <procedure>(z percentile)</procedure> |
---|
274 | returns the inverse of the standard normal distribution. Input is a percentile, between 0.0 and 1.0. |
---|
275 | |
---|
276 | <procedure>( t-distribution degrees-of-freedom percentile)</procedure> |
---|
277 | returns the point in the t-distribution given the {{degrees-of-freedom}} and {{percentile}}. {{degrees-of-freedom}} must be a positive integer, and {{percentile}} a value between 0.0 and 1.0. |
---|
278 | |
---|
279 | <procedure>(chi-square degrees-of-freedom percentile)</procedure> |
---|
280 | returns the point at which chi-square distribution has {{percentile}} to its '''left''', using given {{degrees-of-freedom}}. |
---|
281 | |
---|
282 | <procedure>(chi-square-cdf x degrees-of-freedom)</procedure> |
---|
283 | returns the probability that a random variable is to the '''left''' of {{x}} using the chi-square distribution with given {{degrees-of-freedom}}. |
---|
284 | |
---|
285 | ==== Confidence intervals |
---|
286 | |
---|
287 | These functions report bounds for an observed property of a distribution: the bounds are tighter as the confidence level, alpha, varies from 0.0 to 1.0. |
---|
288 | |
---|
289 | <procedure>(binomial-probability-ci n p alpha)</procedure> |
---|
290 | returns two values, the upper and lower bounds on an observed probability {{p}} from {{n}} trials with confidence {{(1-alpha)}}. |
---|
291 | > (binomial-probability-ci 10 0.8 0.9) |
---|
292 | 0.724273681640625 |
---|
293 | 0.851547241210938 |
---|
294 | ; 2 values |
---|
295 | |
---|
296 | <procedure>(poisson-mu-ci k alpha)</procedure> |
---|
297 | returns two values, the upper and lower bounds on the poisson parameter if {{k}} events are observed; the bound is for confidence {{(1-alpha)}}. |
---|
298 | > (poisson-mu-ci 10 0.9) |
---|
299 | 8.305419921875 |
---|
300 | 10.0635986328125 |
---|
301 | ; 2 values |
---|
302 | |
---|
303 | <procedure>(normal-mean-ci mean standard-deviation k alpha)</procedure> |
---|
304 | returns two values, the upper and lower bounds on the mean of the normal distibution of {{k}} events are observed; the bound is for confidence {{(1-alpha)}}. |
---|
305 | > (normal-mean-ci 0.5 0.1 10 0.8) |
---|
306 | 0.491747852700165 |
---|
307 | 0.508252147299835 |
---|
308 | ; 2 values |
---|
309 | |
---|
310 | <procedure>(normal-mean-ci-on-sequence items alpha)</procedure> |
---|
311 | returns two values, the upper and lower bounds on the mean of the given {{items}}, assuming they are normally distributed; the bound is for confidence {{(1-alpha)}}. |
---|
312 | > (normal-mean-ci-on-sequence '(1 2 3 4 5) 0.9) |
---|
313 | 2.40860081649174 |
---|
314 | 3.59139918350826 |
---|
315 | ; 2 values |
---|
316 | |
---|
317 | <procedure>(normal-variance-ci standard-deviation k alpha)</procedure> |
---|
318 | returns two values, the upper and lower bounds on the variance of the normal distibution of {{k}} events are observed; the bound is for confidence {{(1-alpha)}}. |
---|
319 | |
---|
320 | <procedure>(normal-variance-ci-on-sequence items alpha)</procedure> |
---|
321 | returns two values, the upper and lower bounds on the variance of the given {{items}}, assuming they are normally distributed; the bound is for confidence {{(1-alpha)}}. |
---|
322 | |
---|
323 | <procedure>normal-sd-ci standard-deviation k alpha)</procedure> |
---|
324 | returns two values, the upper and lower bounds on the standard deviation of the normal distibution of {{k}} events are observed; the bound is for confidence {{(1-alpha)}}. |
---|
325 | |
---|
326 | <procedure>(normal-sd-ci-on-sequence sequence items)</procedure> |
---|
327 | returns two values, the upper and lower bounds on the standard deviation of the given {{items}}, assuming they are normally distributed; the bound is for confidence {{(1-alpha)}}. |
---|
328 | |
---|
329 | ==== Hypothesis testing |
---|
330 | |
---|
331 | These functions report on the significance of an observed sample against a given distribution. |
---|
332 | |
---|
333 | ===== (parametric) |
---|
334 | |
---|
335 | <procedure>(z-test x-bar n #:mu #:sigma #:tails)</procedure> |
---|
336 | Given {{x-bar}} the sample mean, {{n}} the number in the sample, {{#:mu}} the distribution mean (defaults to 0), {{#:sigma}} the distribution standard deviation (defaults to 1), and {{#:tails}} the significance to report on: |
---|
337 | |
---|
338 | * {{':both}}, the probability of the difference between {{x-bar}} and {{#:mu}} |
---|
339 | * {{':positive}}, the probability that observation is {{>= x-bar}} |
---|
340 | * {{':negative}}, the probability that observation is {{<= x-bar}} |
---|
341 | |
---|
342 | e.g. given a distribution with mean 50 and standard deviation 10 |
---|
343 | |
---|
344 | ; probability that a single observation is <= 40 |
---|
345 | > (z-test 40 1 #:mu 50 #:sigma 10 #:tails ':negative) |
---|
346 | 0.158655 |
---|
347 | ; probability that 10 observations are <= 40 |
---|
348 | > (z-test 40 10 #:mu 50 #:sigma 10 #:tails ':negative) |
---|
349 | 0.000783 |
---|
350 | ; probability that 5 observations give a mean of 40 |
---|
351 | > (z-test 40 5 #:mu 50 #:sigma 10) |
---|
352 | 0.025347 |
---|
353 | |
---|
354 | <procedure>(z-test-on-sequence observations #:mu #:sigma #:tails)</procedure> |
---|
355 | As for {{z-test}} except {{x-bar}} and {{n}} are computed from given {{observations}}. |
---|
356 | |
---|
357 | <procedure>(t-test-one-sample x-bar sd n mu #:tails)</procedure> |
---|
358 | Given observed data with mean {{x-bar}}, standard devation {{sd}} and number of observations {{n}} ({{n < 30}}), return the significance of the sample compared with the population mean {{mu}}. {{#:tails}} is one of: |
---|
359 | |
---|
360 | * {{':both}} two-sided (default) |
---|
361 | * {{':positive}} one-sided, {{x-bar >= mu}} |
---|
362 | * {{':negative}} one-sided, {{x-bar <= mu}} |
---|
363 | |
---|
364 | <procedure>(t-test-one-sample-on-sequence observations mu #:tails)</procedure> |
---|
365 | As for {{t-test-one-sample}} except {{x-bar}}, {{sd}} and {{n}} are computed from given {{observations}}. |
---|
366 | |
---|
367 | <procedure>(t-test-paired t-bar sd n #:tails)</procedure> |
---|
368 | Computes the significance of the differences between two sequences of data: the differences are given as their mean, {{t-bar}}, standard deviation, {{sd}}, and number of measurements, {{n}}. |
---|
369 | |
---|
370 | <procedure>(t-test-paired-on-sequences before after #:tails)</procedure> |
---|
371 | Computes the significance of the difference between two sequences of data: one before an experimental change and one after. {{#:tails}} is as for {{t-significance}}. |
---|
372 | |
---|
373 | > (t-test-paired-on-sequences '(4 3 5) '(1 1 3)) |
---|
374 | 0.0198039411803931 |
---|
375 | |
---|
376 | <procedure>(t-test-two-sample mean-1 sd-1 n-1 mean-2 sd-2 n-2 #:variances-equal? #:variance-significance-cutoff #:tails)</procedure> |
---|
377 | Computes the significance of the difference of two means given the sample standard deviations and sizes. |
---|
378 | |
---|
379 | <procedure>(t-test-two-sample-on-sequences sequence-1 sequence-2 #:tails)</procedure> |
---|
380 | Significance of difference of two sequences of observations. |
---|
381 | |
---|
382 | <procedure>(f-test variance-1 n1 variance-2 n2 #:tails)</procedure> |
---|
383 | Tests for the equality of two variances. |
---|
384 | |
---|
385 | <procedure>(chi-square-test-one-sample observed-variance sample-size test-variance #:tails)</procedure> |
---|
386 | Tests for significance of difference between an observed and a test variance. |
---|
387 | |
---|
388 | <procedure>(binomial-test-one-sample p-hat n p #:tails #:exact?)</procedure> |
---|
389 | Returns the significance of a one sample test with {{n}} observations, observed probability {{p-hat}} and expected probability {{p}}. |
---|
390 | |
---|
391 | <procedure>(binomial-test-two-sample p-hat-1 n-1 p-hat-2 n-2 #:tails #:exact?)</procedure> |
---|
392 | Returns the significance of a two sample test. |
---|
393 | |
---|
394 | <procedure>(fisher-exact-test a b c d #:tails)</procedure> |
---|
395 | Given a 2x2 contingency table, returns a p value using Fisher's exact test. {{a}} and {{b}} form the first row of the contingency table, {{c}} and {{d}} the second row. |
---|
396 | |
---|
397 | <procedure>(mcnemars-test a-discordant-count b-discordant-count #:exact?)</procedure> |
---|
398 | For measuring effectiveness of, e.g., one treatment over another. {{a-discordant-count}} is the number of times when A worked, {{b-discordant-count}} the number of times B worked. |
---|
399 | |
---|
400 | <procedure>(poisson-test-one-sample observed mu #:tails #:approximate?)</procedure> |
---|
401 | Computes significance of the number of observed events under a Poisson distribution against {{mu}} expected events. |
---|
402 | |
---|
403 | ===== (non parametric) |
---|
404 | |
---|
405 | <procedure>(sign-test plus-count minus-count #:exact? #:tails)</procedure> |
---|
406 | |
---|
407 | <procedure>(sign-test-on-sequence sequence-1 sequence-2 #:exact? #:tails)</procedure> |
---|
408 | Takes two equal-sized sequences of observations, and reports if the entries of one are different to those in the other. |
---|
409 | |
---|
410 | <procedure>(wilcoxon-signed-rank-test differences #:tails)</procedure> |
---|
411 | Given at least 16 differences, reports if the positive differences are significantly larger or smaller than the negative differences. |
---|
412 | |
---|
413 | <procedure>(wilcoxon-signed-rank-test-on-sequences sequence-1 sequence-2 #:tails)</procedure> |
---|
414 | Given two sequences of at least 16 observations, computes {{wilcoxon-signed-rank-test}} on the differences. |
---|
415 | |
---|
416 | <procedure>(chi-square-test-rxc contingency-table)</procedure> |
---|
417 | Given a contingency table (a SRFI-63 array), returns significance of relation between row and column variable. |
---|
418 | |
---|
419 | <procedure>(chi-square-test-for-trend row1-counts row2-counts)</procedure> |
---|
420 | Returns p significance of trend, and prints a string to show if increasing or decreasing. |
---|
421 | |
---|
422 | ==== Sample size estimates |
---|
423 | |
---|
424 | <procedure>(t-test-one-sample-sse mean-1 mean-2 sigma-1 #:alpha #:1-beta #:tails)</procedure> |
---|
425 | Returns the size of sample necessary to distinguish a normally distributed sample with {{mean-2}} from a population {{mean-1}} standard deviation {{sigma-1}}. The significance {{#:alpha}} (defaults to 0.05), power {{#:1-beta}} (0.95) and sides {{#:tails}} (':both) may be altered. |
---|
426 | > (t-test-one-sample-sse 5.0 5.2 0.5) |
---|
427 | 163 |
---|
428 | |
---|
429 | <procedure>(t-test-two-sample-sse mean-1 sigma-1 mean-2 sigma-2 #:alpha #:1-beta #:tails #:sample-ratio)</procedure> |
---|
430 | Returns the size of sample necessary to distinguish a normally distributed sample N(mean-1, sigma-1) from a normally distributed sample N(mean-2, sigma-2). The significance {{#:alpha}} (defaults to 0.05), power {{#:1-beta}} (0.95), sides {{#:tails}} (':both) and sample-ratio {{#:sample-ratio}} (1) may be altered. |
---|
431 | |
---|
432 | <procedure>(t-test-paired-sse difference-mean difference-sigma #:alpha #:1-beta #:tails)</procedure> |
---|
433 | Returns the size of sample to produce a given mean and standard deviation in the differences of two samples. |
---|
434 | |
---|
435 | <procedure>(binomial-test-one-sample-sse p-estimated p-null #:alpha #:1-beta #:tails)</procedure> |
---|
436 | Returns the size of sample needed to test whether an observed probability is significantly different from a particular binomial null hypothesis with a significance alpha and a power 1-beta. |
---|
437 | |
---|
438 | <procedure>(binomial-test-two-sample-sse p-one p-two #:alpha #:1-beta #:tails #:sample-ratio)</procedure> |
---|
439 | Returns the size of sample needed to test if given two binomial probabilities are significantly different. {{#:sample-ratio}} can be given if the two samples differ in size. |
---|
440 | |
---|
441 | <procedure>(binomial-test-paired-sse pd pa #:alpha #:1-beta #:tails)</procedure> |
---|
442 | Sample size estimate for McNemar's discordant pairs test. |
---|
443 | |
---|
444 | <procedure>(correlation-sse rho #:alpha #:1-beta)</procedure> |
---|
445 | Returns the size of sample necessary to find a correlation of value {{rho}} with significance {{#:alpha}} (defaults to 0.05) and power {{#:1-beta}} (defaults to 0.95). |
---|
446 | > (correlation-sse 0.80 #:alpha 0.05 #:1-beta 0.9) |
---|
447 | 11 |
---|
448 | |
---|
449 | ==== Correlation and regression |
---|
450 | |
---|
451 | <procedure>(linear-regression xs ys)</procedure> |
---|
452 | |
---|
453 | Given a line definition as lists of point coordinates, first prints to |
---|
454 | the terminal and then returns 5 '''values''' for the best fitting line |
---|
455 | through the points: |
---|
456 | |
---|
457 | * the y-intercept |
---|
458 | * the slope |
---|
459 | * the correlation coefficient, r |
---|
460 | * the square of the correlation coefficient, r^2 |
---|
461 | * the significance of the difference of the slope from zero, p |
---|
462 | |
---|
463 | (This is also called the Pearson correlation; used when relation expected to be linear. Also see {{spearman-rank-correlation}}.) |
---|
464 | |
---|
465 | > (linear-regression '(1.0 2.0 3.0) '(0.1 0.3 0.8)) |
---|
466 | Intercept = -0.3, slope = 0.35, r = 0.970725343394151, R^2 = 0.942307692307692, p = 0.154420958311267 |
---|
467 | -0.3 |
---|
468 | 0.35 |
---|
469 | 0.970725343394151 |
---|
470 | 0.942307692307692 |
---|
471 | 0.154420958311267 |
---|
472 | ; 5 values |
---|
473 | |
---|
474 | <procedure>(correlation-coefficient xs ys)</procedure> |
---|
475 | As above, but only returns the value of ''r'': |
---|
476 | |
---|
477 | > (correlation-coefficient '(1.0 2.0 3.0) '(0.1 0.3 0.8)) |
---|
478 | 0.970725343394151 |
---|
479 | |
---|
480 | <procedure>(correlation-test-two-sample r1 n1 r2 n2 #:tails)</procedure> |
---|
481 | Returns the significance of the similarity between two correlations. {{#:tails}} determines how the comparison is made: {{':both}} measures the difference, {{':negative}} if {{r1 < r2}} and {{#':positive}} if {{r2 > r1}}. |
---|
482 | |
---|
483 | <procedure>(correlation-test-two-sample-on-sequences points-1 points-2 #:tails)</procedure> |
---|
484 | As above, but computes the correlations from given lists of points. |
---|
485 | |
---|
486 | <procedure>(spearman-rank-correlation xs ys)</procedure> |
---|
487 | Returns two '''values''', the Spearman Rank measure of correlation between the given lists of point coordinates, and the p-significance of the correlation. (This correlation is used for non-linear relations; compare with {{linear-regression}}.) |
---|
488 | |
---|
489 | ==== Significance test functions |
---|
490 | |
---|
491 | <procedure>(t-significance t-value degrees-of-freedom #:tails)</procedure> |
---|
492 | returns the probability of {{t-value}} for given {{degrees-of-freedom}}. The keyword {{#:tails}} modifies the calculation to be two-sided (the default) with {{':both}}, or one-sided, {{':positive}} or {{':negative}}. |
---|
493 | |
---|
494 | > (t-significance 0.2 5) |
---|
495 | 0.849360513995829 |
---|
496 | > (t-significance 0.2 5 #:tails ':positive) |
---|
497 | 0.424680256997915 |
---|
498 | > (t-significance 0.2 5 #:tails ':negative) |
---|
499 | 0.575319743002086 |
---|
500 | |
---|
501 | <procedure>(f-significance f-value numerator-dof denominator-dof #:one-tailed?)</procedure> |
---|
502 | returns the probability of {{f-value}} for given {{numerator-dof}} and {{denominator-dof}}. The boolean keyword {{#:one-tailed?}} indicates if calculation is two-sided (the default) or not. |
---|
503 | |
---|
504 | > (f-significance 1.5 8 2) |
---|
505 | 0.920449812578091 |
---|
506 | > (f-significance 1.5 8 2 #:one-tailed? #t) |
---|
507 | 0.460224906289046 |
---|
508 | |
---|
509 | |
---|
510 | === Authors |
---|
511 | |
---|
512 | [[/users/peter-lane|Peter Lane]] wrote the Scheme version of this library. The original Lisp version was written by [[http://compbio.ucdenver.edu/hunter/|Larry Hunter]]. |
---|
513 | |
---|
514 | === License |
---|
515 | |
---|
516 | GPL version 3.0. |
---|
517 | |
---|
518 | === Requirements |
---|
519 | |
---|
520 | Needs srfi-1, srfi-25, srfi-63, srfi-69, vector-lib, foreign, format |
---|
521 | |
---|
522 | Uses the GNU scientific library for basic numeric processing, so requires libgsl, libgslcblas and the development files for libgsl. |
---|
523 | |
---|
524 | === Version History |
---|
525 | |
---|
526 | * 0.11: refactoring correlation and regression interface to take two separate dataset arguments |
---|
527 | * 0.9: ported to CHICKEN 5 |
---|
528 | * 0.8: added cumsum and random-weighted-sample |
---|
529 | * 0.5: fixed warning in compilation (thanks to Felix for pointing it out) |
---|
530 | * 0.4: all functions should now be working |
---|
531 | * 0.3: some error fixes and addition of tests for majority of functions |
---|
532 | * 0.2: fixed some errors in keywords and find-critical-value |
---|
533 | * 0.1: initial package |
---|