StatisticsTest.h 9.42 KB
Newer Older
1
2
3
4
5
6
// Mantid Repository : https://github.com/mantidproject/mantid
//
// Copyright © 2018 ISIS Rutherford Appleton Laboratory UKRI,
//     NScD Oak Ridge National Laboratory, European Spallation Source
//     & Institut Laue - Langevin
// SPDX - License - Identifier: GPL - 3.0 +
7
8
9
#ifndef STATISTICSTEST_H_
#define STATISTICSTEST_H_

10
11
#include "MantidKernel/Statistics.h"
#include <algorithm>
12
#include <cmath>
LamarMoore's avatar
LamarMoore committed
13
#include <cxxtest/TestSuite.h>
14
#include <string>
LamarMoore's avatar
LamarMoore committed
15
#include <vector>
16
17
18
19
20

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

21
class StatisticsTest : public CxxTest::TestSuite {
22
public:
23
  void test_Doubles_And_Default_Flags_Calculates_All_Stats() {
24
25
26
27
28
29
30
31
32
33
34
35
36
37
    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);
38
39
  }

40
  void test_Doubles_With_Sorted_Data() {
41
42
43
44
45
46
47
48
    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());

49
50
    Statistics stats =
        getStatistics(data, (StatOptions::Median | StatOptions::SortedData));
51

52
53
54
55
    TS_ASSERT(std::isnan(stats.mean));
    TS_ASSERT(std::isnan(stats.standard_deviation));
    TS_ASSERT(std::isnan(stats.minimum));
    TS_ASSERT(std::isnan(stats.maximum));
56
57
58
    TS_ASSERT_EQUALS(stats.median, 17.2);
  }

59
60
  void
  test_Unsorted_Data_With_Sorted_Flag_Gives_Expected_Incorrect_Result_For_Median() {
61
62
63
64
65
66
67
    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);

68
69
    Statistics stats =
        getStatistics(data, (StatOptions::Median | StatOptions::SortedData));
70

71
72
73
74
    TS_ASSERT(std::isnan(stats.mean));
    TS_ASSERT(std::isnan(stats.standard_deviation));
    TS_ASSERT(std::isnan(stats.minimum));
    TS_ASSERT(std::isnan(stats.maximum));
75
76
    TS_ASSERT_EQUALS(stats.median, 16.5);
  }
77
78

  void test_Doubles_With_Corrected_StdDev_Calculates_Mean() {
79
80
81
82
83
84
85
86
87
88
89
90
91
92
    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);
93
    TS_ASSERT(std::isnan(stats.median));
94
95
96
97
98
99
100
101
102
  }

  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
103

104
105
    Statistics justMean = getStatistics(data, StatOptions::Mean);
    TS_ASSERT_EQUALS(justMean.mean, 16.54);
106
107
108
109
    TS_ASSERT(std::isnan(justMean.standard_deviation));
    TS_ASSERT(std::isnan(justMean.minimum));
    TS_ASSERT(std::isnan(justMean.maximum));
    TS_ASSERT(std::isnan(justMean.median));
Lynch, Vickie's avatar
Lynch, Vickie committed
110
  }
111
112

  void testZscores() {
Lynch, Vickie's avatar
Lynch, Vickie committed
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
    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);
141
142
  }

143
  void testDoubleSingle() {
144
145
146
147
148
149
150
151
152
153
154
155
    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.);
  }

156
  void testInt32Even() {
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
    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);
  }

174
  bool my_isnan(const double number) { return number != number; }
175

176
  void testString() {
177
    vector<string> data{"hi there"};
178
179
180
181
182
183
184
185
186

    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));
  }
187
188

  /** Test function to calculate Rwp
LamarMoore's avatar
LamarMoore committed
189
   */
190
  void testRwp() {
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
    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;

211
    Rfactor rfactor = getRFactor(obsY, calY, obsE);
212

213
    TS_ASSERT_DELTA(rfactor.Rwp, 0.1582, 0.0001);
214
215
216
  }

  /** Test throw exception
LamarMoore's avatar
LamarMoore committed
217
   */
218
  void testRwpException1() {
219
220
221
    vector<double> obsY{1.0, 2.0, 3.0, 1.0};
    vector<double> calY{1.1, 2.1, 3.5, 1.3};
    vector<double> obsE{1.0, 1.2, 1.4};
222
223

    TS_ASSERT_THROWS_ANYTHING(getRFactor(obsY, calY, obsE));
224
  }
225
226

  /** Test throw exception on empty array
LamarMoore's avatar
LamarMoore committed
227
   */
228
  void testRwpException2() {
229
230
231
232
233
234
235
    vector<double> obsY;
    vector<double> calY;
    vector<double> obsE;

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

236
  /// Test moment calculations about origin and mean
237
  void test_getMoments() {
238
239
240
241
242
243
244
    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
245
    double expVar = mean * mean + sigma * sigma;
246
    // skew about origin
247
    double expSkew = mean * mean * mean + 3. * mean * sigma * sigma;
248
249
250
251
252
253
254
255
256

    // 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;

257
    TS_ASSERT_THROWS(getMomentsAboutOrigin(x, y), const std::out_of_range &);
258
259

    // now calculate the y-values
260
261
262
    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)));
263
264
265
266
    }

    // Normal distribution values are taken from the wikipedia page
    {
267
      std::cout << "Normal distribution about origin\n";
268
269
270
271
      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);
272
273
      TS_ASSERT_DELTA(aboutOrigin[2], expVar, .001 * expVar);
      TS_ASSERT_DELTA(aboutOrigin[3], expSkew, .001 * expSkew);
274

275
      std::cout << "Normal distribution about mean\n";
276
277
278
279
      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);
280
281
      TS_ASSERT_DELTA(aboutMean[2], sigma * sigma, .001 * expVar);
      TS_ASSERT_DELTA(aboutMean[3], 0., .0001 * expSkew);
282
283
284
285
    }

    // Now a gaussian function as a histogram
    y.clear();
286
287
    for (size_t i = 0; i < numX - 1;
         ++i) // one less y than x makes it a histogram
288
    {
289
290
291
292
293
      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));
294
      //      std::cout << i << ":\t" << x[i] << "\t" << y[i] << '\n';
295
296
297
298
    }

    // Normal distribution values are taken from the wikipedia page
    {
299
      std::cout << "Normal distribution about origin\n";
300
301
302
303
      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);
304
305
      TS_ASSERT_DELTA(aboutOrigin[2], expVar, .001 * expVar);
      TS_ASSERT_DELTA(aboutOrigin[3], expSkew, .001 * expSkew);
306

307
      std::cout << "Normal distribution about mean\n";
308
309
310
311
      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);
312
313
      TS_ASSERT_DELTA(aboutMean[2], sigma * sigma, .001 * expVar);
      TS_ASSERT_DELTA(aboutMean[3], 0., .0001 * expSkew);
314
315
    }
  }
316
317
318
};

#endif // STATISTICSTEST_H_