StatisticsTest.h 9.53 KB
Newer Older
1
2
3
#ifndef STATISTICSTEST_H_
#define STATISTICSTEST_H_

4
5
#include "MantidKernel/Statistics.h"
#include <boost/math/special_functions/fpclassify.hpp>
6
#include <cxxtest/TestSuite.h>
7
#include <algorithm>
8
#include <cmath>
9
10
11
12
13
14
15
#include <vector>
#include <string>

using namespace Mantid::Kernel;
using std::string;
using std::vector;

16
class StatisticsTest : public CxxTest::TestSuite {
17
public:
18
  void test_Doubles_And_Default_Flags_Calculates_All_Stats() {
19
20
21
22
23
24
25
26
27
28
29
30
31
32
    vector<double> data;
    data.push_back(17.2);
    data.push_back(18.1);
    data.push_back(16.5);
    data.push_back(18.3);
    data.push_back(12.6);

    Statistics stats = getStatistics(data);

    TS_ASSERT_EQUALS(stats.mean, 16.54);
    TS_ASSERT_DELTA(stats.standard_deviation, 2.0732, 0.0001);
    TS_ASSERT_EQUALS(stats.minimum, 12.6);
    TS_ASSERT_EQUALS(stats.maximum, 18.3);
    TS_ASSERT_EQUALS(stats.median, 17.2);
33
34
  }

35
  void test_Doubles_With_Sorted_Data() {
36
37
38
39
40
41
42
43
    vector<double> data;
    data.push_back(17.2);
    data.push_back(18.1);
    data.push_back(16.5);
    data.push_back(18.3);
    data.push_back(12.6);
    sort(data.begin(), data.end());

44
45
    Statistics stats =
        getStatistics(data, (StatOptions::Median | StatOptions::SortedData));
46
47
48
49
50
51
52
53

    TS_ASSERT(boost::math::isnan(stats.mean));
    TS_ASSERT(boost::math::isnan(stats.standard_deviation));
    TS_ASSERT(boost::math::isnan(stats.minimum));
    TS_ASSERT(boost::math::isnan(stats.maximum));
    TS_ASSERT_EQUALS(stats.median, 17.2);
  }

54
55
  void
  test_Unsorted_Data_With_Sorted_Flag_Gives_Expected_Incorrect_Result_For_Median() {
56
57
58
59
60
61
62
    vector<double> data;
    data.push_back(17.2);
    data.push_back(18.1);
    data.push_back(16.5);
    data.push_back(18.3);
    data.push_back(12.6);

63
64
    Statistics stats =
        getStatistics(data, (StatOptions::Median | StatOptions::SortedData));
65
66
67
68
69
70
71

    TS_ASSERT(boost::math::isnan(stats.mean));
    TS_ASSERT(boost::math::isnan(stats.standard_deviation));
    TS_ASSERT(boost::math::isnan(stats.minimum));
    TS_ASSERT(boost::math::isnan(stats.maximum));
    TS_ASSERT_EQUALS(stats.median, 16.5);
  }
72
73

  void test_Doubles_With_Corrected_StdDev_Calculates_Mean() {
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
    vector<double> data;
    data.push_back(17.2);
    data.push_back(18.1);
    data.push_back(16.5);
    data.push_back(18.3);
    data.push_back(12.6);
    sort(data.begin(), data.end());

    Statistics stats = getStatistics(data, StatOptions::CorrectedStdDev);

    TS_ASSERT_EQUALS(stats.mean, 16.54);
    TS_ASSERT_DELTA(stats.standard_deviation, 2.3179, 0.0001);
    TS_ASSERT_EQUALS(stats.minimum, 12.6);
    TS_ASSERT_EQUALS(stats.maximum, 18.3);
    TS_ASSERT(boost::math::isnan(stats.median));
  }

  void test_Types_Can_Be_Disabled_With_Flags() {
    vector<double> data;
    data.push_back(17.2);
    data.push_back(18.1);
    data.push_back(16.5);
    data.push_back(18.3);
    data.push_back(12.6);
Lynch, Vickie's avatar
Lynch, Vickie committed
98

99
100
101
102
103
104
    Statistics justMean = getStatistics(data, StatOptions::Mean);
    TS_ASSERT_EQUALS(justMean.mean, 16.54);
    TS_ASSERT(boost::math::isnan(justMean.standard_deviation));
    TS_ASSERT(boost::math::isnan(justMean.minimum));
    TS_ASSERT(boost::math::isnan(justMean.maximum));
    TS_ASSERT(boost::math::isnan(justMean.median));
Lynch, Vickie's avatar
Lynch, Vickie committed
105
  }
106
107

  void testZscores() {
Lynch, Vickie's avatar
Lynch, Vickie committed
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
    vector<double> data;
    data.push_back(12);
    data.push_back(13);
    data.push_back(9);
    data.push_back(18);
    data.push_back(7);
    data.push_back(9);
    data.push_back(14);
    data.push_back(16);
    data.push_back(10);
    data.push_back(12);
    data.push_back(7);
    data.push_back(13);
    data.push_back(14);
    data.push_back(19);
    data.push_back(10);
    data.push_back(16);
    data.push_back(12);
    data.push_back(16);
    data.push_back(19);
    data.push_back(11);

    std::vector<double> Zscore = getZscore(data);
    TS_ASSERT_DELTA(Zscore[4], 1.6397, 0.0001);
    TS_ASSERT_DELTA(Zscore[6], 0.3223, 0.0001);
    std::vector<double> ZModscore = getModifiedZscore(data);
    TS_ASSERT_DELTA(ZModscore[4], 1.2365, 0.0001);
    TS_ASSERT_DELTA(ZModscore[6], 0.3372, 0.0001);
136
137
  }

138
  void testDoubleSingle() {
139
140
141
142
143
144
145
146
147
148
149
150
    vector<double> data;
    data.push_back(42.);

    Statistics stats = getStatistics(data);

    TS_ASSERT_EQUALS(stats.mean, 42.);
    TS_ASSERT_EQUALS(stats.standard_deviation, 0.);
    TS_ASSERT_EQUALS(stats.minimum, 42.);
    TS_ASSERT_EQUALS(stats.maximum, 42.);
    TS_ASSERT_EQUALS(stats.median, 42.);
  }

151
  void testInt32Even() {
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
    vector<int32_t> data;
    data.push_back(1);
    data.push_back(2);
    data.push_back(3);
    data.push_back(4);
    data.push_back(5);
    data.push_back(6);

    Statistics stats = getStatistics(data);

    TS_ASSERT_EQUALS(stats.mean, 3.5);
    TS_ASSERT_DELTA(stats.standard_deviation, 1.7078, 0.0001);
    TS_ASSERT_EQUALS(stats.minimum, 1.);
    TS_ASSERT_EQUALS(stats.maximum, 6.);
    TS_ASSERT_EQUALS(stats.median, 3.5);
  }

169
  bool my_isnan(const double number) { return number != number; }
170

171
  void testString() {
172
    vector<string> data;
173
    data.emplace_back("hi there");
174
175
176
177
178
179
180
181
182

    Statistics stats = getStatistics(data);

    TS_ASSERT(my_isnan(stats.mean));
    TS_ASSERT(my_isnan(stats.standard_deviation));
    TS_ASSERT(my_isnan(stats.minimum));
    TS_ASSERT(my_isnan(stats.maximum));
    TS_ASSERT(my_isnan(stats.median));
  }
183
184
185

  /** Test function to calculate Rwp
    */
186
  void testRwp() {
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
    vector<double> obsY(4);
    vector<double> calY(4);
    vector<double> obsE(4);

    obsY[0] = 1.0;
    calY[0] = 1.1;
    obsE[0] = 1.0;

    obsY[1] = 2.0;
    calY[1] = 2.1;
    obsE[1] = 1.2;

    obsY[2] = 3.0;
    calY[2] = 3.5;
    obsE[2] = 1.4;

    obsY[3] = 1.0;
    calY[3] = 1.3;
    obsE[3] = 1.0;

207
    Rfactor rfactor = getRFactor(obsY, calY, obsE);
208

209
    TS_ASSERT_DELTA(rfactor.Rwp, 0.1582, 0.0001);
210
211
212
213
  }

  /** Test throw exception
    */
214
  void testRwpException1() {
215
216
217
218
219
220
221
222
223
224
225
    vector<double> obsY(4);
    vector<double> calY(4);
    vector<double> obsE(3);

    obsY[0] = 1.0;
    calY[0] = 1.1;
    obsE[0] = 1.0;

    obsY[1] = 2.0;
    calY[1] = 2.1;
    obsE[1] = 1.2;
226

227
228
229
230
231
232
233
234
    obsY[2] = 3.0;
    calY[2] = 3.5;
    obsE[2] = 1.4;

    obsY[3] = 1.0;
    calY[3] = 1.3;

    TS_ASSERT_THROWS_ANYTHING(getRFactor(obsY, calY, obsE));
235
  }
236
237
238

  /** Test throw exception on empty array
    */
239
  void testRwpException2() {
240
241
242
243
244
245
246
    vector<double> obsY;
    vector<double> calY;
    vector<double> obsE;

    TS_ASSERT_THROWS_ANYTHING(getRFactor(obsY, calY, obsE));
  }

247
  /// Test moment calculations about origin and mean
248
  void test_getMoments() {
249
250
251
252
253
254
255
    const double mean = 5.;
    const double sigma = 4.;
    const double deltaX = .2;
    const size_t numX = 200;
    // calculate to have same number of points left and right of function
    const double offsetX = mean - (.5 * deltaX * static_cast<double>(numX));
    // variance about origin
256
    double expVar = mean * mean + sigma * sigma;
257
    // skew about origin
258
    double expSkew = mean * mean * mean + 3. * mean * sigma * sigma;
259
260
261
262
263
264
265
266
267
268
269
270

    // x-values to try out
    vector<double> x;
    for (size_t i = 0; i < numX; ++i)
      x.push_back(static_cast<double>(i) * deltaX + offsetX);

    // just declare so we can have test of exception handling
    vector<double> y;

    TS_ASSERT_THROWS(getMomentsAboutOrigin(x, y), std::out_of_range);

    // now calculate the y-values
271
272
273
    for (size_t i = 0; i < numX; ++i) {
      double temp = (x[i] - mean) / sigma;
      y.push_back(exp(-.5 * temp * temp) / (sigma * sqrt(2. * M_PI)));
274
275
276
277
278
279
280
281
282
    }

    // Normal distribution values are taken from the wikipedia page
    {
      std::cout << "Normal distribution about origin" << std::endl;
      vector<double> aboutOrigin = getMomentsAboutOrigin(x, y);
      TS_ASSERT_EQUALS(aboutOrigin.size(), 4);
      TS_ASSERT_DELTA(aboutOrigin[0], 1., .0001);
      TS_ASSERT_DELTA(aboutOrigin[1], mean, .0001);
283
284
      TS_ASSERT_DELTA(aboutOrigin[2], expVar, .001 * expVar);
      TS_ASSERT_DELTA(aboutOrigin[3], expSkew, .001 * expSkew);
285
286
287
288
289
290

      std::cout << "Normal distribution about mean" << std::endl;
      vector<double> aboutMean = getMomentsAboutMean(x, y);
      TS_ASSERT_EQUALS(aboutMean.size(), 4);
      TS_ASSERT_DELTA(aboutMean[0], 1., .0001);
      TS_ASSERT_DELTA(aboutMean[1], 0., .0001);
291
292
      TS_ASSERT_DELTA(aboutMean[2], sigma * sigma, .001 * expVar);
      TS_ASSERT_DELTA(aboutMean[3], 0., .0001 * expSkew);
293
294
295
296
    }

    // Now a gaussian function as a histogram
    y.clear();
297
298
    for (size_t i = 0; i < numX - 1;
         ++i) // one less y than x makes it a histogram
299
    {
300
301
302
303
304
305
      double templeft = (x[i] - mean) / sigma;
      templeft = exp(-.5 * templeft * templeft) / (sigma * sqrt(2. * M_PI));
      double tempright = (x[i + 1] - mean) / sigma;
      tempright = exp(-.5 * tempright * tempright) / (sigma * sqrt(2. * M_PI));
      y.push_back(.5 * deltaX * (templeft + tempright));
      //      std::cout << i << ":\t" << x[i] << "\t" << y[i] << std::endl;
306
307
308
309
310
311
312
313
314
    }

    // Normal distribution values are taken from the wikipedia page
    {
      std::cout << "Normal distribution about origin" << std::endl;
      vector<double> aboutOrigin = getMomentsAboutOrigin(x, y);
      TS_ASSERT_EQUALS(aboutOrigin.size(), 4);
      TS_ASSERT_DELTA(aboutOrigin[0], 1., .0001);
      TS_ASSERT_DELTA(aboutOrigin[1], mean, .0001);
315
316
      TS_ASSERT_DELTA(aboutOrigin[2], expVar, .001 * expVar);
      TS_ASSERT_DELTA(aboutOrigin[3], expSkew, .001 * expSkew);
317
318
319
320
321
322

      std::cout << "Normal distribution about mean" << std::endl;
      vector<double> aboutMean = getMomentsAboutMean(x, y);
      TS_ASSERT_EQUALS(aboutMean.size(), 4);
      TS_ASSERT_DELTA(aboutMean[0], 1., .0001);
      TS_ASSERT_DELTA(aboutMean[1], 0., .0001);
323
324
      TS_ASSERT_DELTA(aboutMean[2], sigma * sigma, .001 * expVar);
      TS_ASSERT_DELTA(aboutMean[3], 0., .0001 * expSkew);
325
326
    }
  }
327
328
329
};

#endif // STATISTICSTEST_H_