pintos-os.org Git - pspp/blob - tests/language/stats/examine.at

   1 dnl PSPP - a program for statistical analysis.
   2 dnl Copyright (C) 2017, 2019 Free Software Foundation, Inc.
   3 dnl
   4 dnl This program is free software: you can redistribute it and/or modify
   5 dnl it under the terms of the GNU General Public License as published by
   6 dnl the Free Software Foundation, either version 3 of the License, or
   7 dnl (at your option) any later version.
   8 dnl
   9 dnl This program is distributed in the hope that it will be useful,
  10 dnl but WITHOUT ANY WARRANTY; without even the implied warranty of
  11 dnl MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12 dnl GNU General Public License for more details.
  13 dnl
  14 dnl You should have received a copy of the GNU General Public License
  15 dnl along with this program.  If not, see <http://www.gnu.org/licenses/>.
  16 dnl
  17 AT_BANNER([EXAMINE])
  18
  19 AT_SETUP([EXAMINE])
  20 AT_KEYWORDS([categorical categoricals])
  21 AT_DATA([examine.sps], [
  22 DATA LIST LIST /QUALITY * W * BRAND * .
  23 BEGIN DATA
  24 3  1  1
  25 2  2  1
  26 1  2  1
  27 1  1  1
  28 4  1  1
  29 4  1  1
  30 5  1  2
  31 2  1  2
  32 4  4  2
  33 2  1  2
  34 3  1  2
  35 7  1  3
  36 4  2  3
  37 5  3  3
  38 3  1  3
  39 6  1  3
  40 END DATA
  41
  42 WEIGHT BY w.
  43
  44 VARIABLE LABELS brand   'Manufacturer'.
  45 VARIABLE LABELS quality 'Breaking Strain'.
  46
  47 VALUE LABELS /brand 1 'Aspeger' 2 'Bloggs' 3 'Charlies'.
  48
  49 LIST /FORMAT=NUMBERED.
  50
  51 EXAMINE
  52         quality BY brand
  53         /STATISTICS descriptives extreme(3)
  54         .
  55 ])
  56
  57
  58 dnl In the following data, only the extreme values have been checked.
  59 dnl The descriptives have been blindly pasted.
  60 AT_CHECK([pspp -O format=csv examine.sps], [0], [dnl
  61 Table: Reading free-form data from INLINE.
  62 Variable,Format
  63 QUALITY,F8.0
  64 W,F8.0
  65 BRAND,F8.0
  66
  67 Table: Data List
  68 Case Number,QUALITY,W,BRAND
  69 1,3.00,1.00,1.00
  70 2,2.00,2.00,1.00
  71 3,1.00,2.00,1.00
  72 4,1.00,1.00,1.00
  73 5,4.00,1.00,1.00
  74 6,4.00,1.00,1.00
  75 7,5.00,1.00,2.00
  76 8,2.00,1.00,2.00
  77 9,4.00,4.00,2.00
  78 10,2.00,1.00,2.00
  79 11,3.00,1.00,2.00
  80 12,7.00,1.00,3.00
  81 13,4.00,2.00,3.00
  82 14,5.00,3.00,3.00
  83 15,3.00,1.00,3.00
  84 16,6.00,1.00,3.00
  85
  86 Table: Case Processing Summary
  87 ,Cases,,,,,
  88 ,Valid,,Missing,,Total,
  89 ,N,Percent,N,Percent,N,Percent
  90 Breaking Strain,24.00,100.0%,.00,.0%,24.00,100.0%
  91
  92 Table: Extreme Values
  93 ,,,Case Number,Value
  94 Breaking Strain,Highest,1,12,7.00
  95 ,,2,16,6.00
  96 ,,3,14,5.00
  97 ,Lowest,1,3,1.00
  98 ,,2,4,1.00
  99 ,,3,2,2.00
 100
 101 Table: Descriptives
 102 ,,,Statistic,Std. Error
 103 Breaking Strain,Mean,,3.54,.32
 104 ,95% Confidence Interval for Mean,Lower Bound,2.87,
 105 ,,Upper Bound,4.21,
 106 ,5% Trimmed Mean,,3.50,
 107 ,Median,,4.00,
 108 ,Variance,,2.52,
 109 ,Std. Deviation,,1.59,
 110 ,Minimum,,1.00,
 111 ,Maximum,,7.00,
 112 ,Range,,6.00,
 113 ,Interquartile Range,,2.75,
 114 ,Skewness,,.06,.47
 115 ,Kurtosis,,-.36,.92
 116
 117 Table: Case Processing Summary
 118 ,Manufacturer,Cases,,,,,
 119 ,,Valid,,Missing,,Total,
 120 ,,N,Percent,N,Percent,N,Percent
 121 Breaking Strain,Aspeger,8.00,100.0%,.00,.0%,8.00,100.0%
 122 ,Bloggs,8.00,100.0%,.00,.0%,8.00,100.0%
 123 ,Charlies,8.00,100.0%,.00,.0%,8.00,100.0%
 124
 125 Table: Extreme Values
 126 ,Manufacturer,,,Case Number,Value
 127 Breaking Strain,Aspeger,Highest,1,6,4.00
 128 ,,,2,5,4.00
 129 ,,,3,1,3.00
 130 ,,Lowest,1,3,1.00
 131 ,,,2,4,1.00
 132 ,,,3,2,2.00
 133 ,Bloggs,Highest,1,7,5.00
 134 ,,,2,9,4.00
 135 ,,,3,11,3.00
 136 ,,Lowest,1,8,2.00
 137 ,,,2,10,2.00
 138 ,,,3,11,3.00
 139 ,Charlies,Highest,1,12,7.00
 140 ,,,2,16,6.00
 141 ,,,3,14,5.00
 142 ,,Lowest,1,15,3.00
 143 ,,,2,13,4.00
 144 ,,,3,14,5.00
 145
 146 Table: Descriptives
 147 ,Manufacturer,,,Statistic,Std. Error
 148 Breaking Strain,Aspeger,Mean,,2.25,.45
 149 ,,95% Confidence Interval for Mean,Lower Bound,1.18,
 150 ,,,Upper Bound,3.32,
 151 ,,5% Trimmed Mean,,2.22,
 152 ,,Median,,2.00,
 153 ,,Variance,,1.64,
 154 ,,Std. Deviation,,1.28,
 155 ,,Minimum,,1.00,
 156 ,,Maximum,,4.00,
 157 ,,Range,,3.00,
 158 ,,Interquartile Range,,2.75,
 159 ,,Skewness,,.47,.75
 160 ,,Kurtosis,,-1.55,1.48
 161 ,Bloggs,Mean,,3.50,.38
 162 ,,95% Confidence Interval for Mean,Lower Bound,2.61,
 163 ,,,Upper Bound,4.39,
 164 ,,5% Trimmed Mean,,3.50,
 165 ,,Median,,4.00,
 166 ,,Variance,,1.14,
 167 ,,Std. Deviation,,1.07,
 168 ,,Minimum,,2.00,
 169 ,,Maximum,,5.00,
 170 ,,Range,,3.00,
 171 ,,Interquartile Range,,1.75,
 172 ,,Skewness,,-.47,.75
 173 ,,Kurtosis,,-.83,1.48
 174 ,Charlies,Mean,,4.88,.44
 175 ,,95% Confidence Interval for Mean,Lower Bound,3.83,
 176 ,,,Upper Bound,5.92,
 177 ,,5% Trimmed Mean,,4.86,
 178 ,,Median,,5.00,
 179 ,,Variance,,1.55,
 180 ,,Std. Deviation,,1.25,
 181 ,,Minimum,,3.00,
 182 ,,Maximum,,7.00,
 183 ,,Range,,4.00,
 184 ,,Interquartile Range,,1.75,
 185 ,,Skewness,,.30,.75
 186 ,,Kurtosis,,.15,1.48
 187 ])
 188
 189 AT_CLEANUP
 190
 191 AT_SETUP([EXAMINE -- extremes])
 192 AT_KEYWORDS([categorical categoricals])
 193 AT_DATA([examine.sps], [dnl
 194 data list free /V1 W
 195 begin data.
 196 1  1
 197 2  1
 198 3  2
 199 3  1
 200 4  1
 201 5  1
 202 6  1
 203 7  1
 204 8  1
 205 9  1
 206 10 1
 207 11 1
 208 12 1
 209 13 1
 210 14 1
 211 15 1
 212 16 1
 213 17 1
 214 18 2
 215 19 1
 216 20 1
 217 end data.
 218
 219 weight by w.
 220
 221 examine v1
 222  /statistics=extreme(6)
 223  .
 224 ])
 225
 226 AT_CHECK([pspp -O format=csv examine.sps], [0],[dnl
 227 Table: Case Processing Summary
 228 ,Cases,,,,,
 229 ,Valid,,Missing,,Total,
 230 ,N,Percent,N,Percent,N,Percent
 231 V1,23.00,100.0%,.00,.0%,23.00,100.0%
 232
 233 Table: Extreme Values
 234 ,,,Case Number,Value
 235 V1,Highest,1,21,20.00
 236 ,,2,20,19.00
 237 ,,3,19,18.00
 238 ,,4,18,17.00
 239 ,,5,17,16.00
 240 ,,6,16,15.00
 241 ,Lowest,1,1,1.00
 242 ,,2,2,2.00
 243 ,,3,3,3.00
 244 ,,4,4,3.00
 245 ,,5,5,4.00
 246 ,,6,6,5.00
 247 ])
 248
 249 AT_CLEANUP
 250
 251
 252 AT_SETUP([EXAMINE -- extremes with fractional weights])
 253 AT_KEYWORDS([categorical categoricals])
 254 AT_DATA([extreme.sps], [dnl
 255 set format=F20.3.
 256 data list notable list /w * x *.
 257 begin data.
 258  0.88  300000
 259  0.86  320000
 260  0.98  480000
 261  0.93  960000
 262  1.35  960000
 263  1.31  960000
 264  0.88  960000
 265  0.88  1080000
 266  0.88  1080000
 267  0.95  1200000
 268  1.47  1200000
 269  0.93  1200000
 270  0.98  1320000
 271  1.31  1380000
 272  0.93  1440000
 273  0.88  1560000
 274  1.56  1560000
 275  1.47  1560000
 276 end data.
 277
 278 weight by w.
 279
 280
 281 EXAMINE
 282         x
 283         /STATISTICS = DESCRIPTIVES EXTREME (5)
 284         .
 285 ])
 286
 287 AT_CHECK([pspp -O format=csv  extreme.sps], [0], [dnl
 288 Table: Case Processing Summary
 289 ,Cases,,,,,
 290 ,Valid,,Missing,,Total,
 291 ,N,Percent,N,Percent,N,Percent
 292 x,19.430,100.0%,.000,.0%,19.430,100.0%
 293
 294 Table: Extreme Values
 295 ,,,Case Number,Value
 296 x,Highest,1,18,1560000.000
 297 ,,2,17,1560000.000
 298 ,,3,16,1560000.000
 299 ,,4,15,1440000.000
 300 ,,5,14,1380000.000
 301 ,Lowest,1,1,300000.000
 302 ,,2,2,320000.000
 303 ,,3,3,480000.000
 304 ,,4,4,960000.000
 305 ,,5,5,960000.000
 306
 307 Table: Descriptives
 308 ,,,Statistic,Std. Error
 309 x,Mean,,1120010.293,86222.178
 310 ,95% Confidence Interval for Mean,Lower Bound,939166.693,
 311 ,,Upper Bound,1300853.894,
 312 ,5% Trimmed Mean,,1141017.899,
 313 ,Median,,1200000.000,
 314 ,Variance,,144447748124.869,
 315 ,Std. Deviation,,380062.821,
 316 ,Minimum,,300000.000,
 317 ,Maximum,,1560000.000,
 318 ,Range,,1260000.000,
 319 ,Interquartile Range,,467258.065,
 320 ,Skewness,,-.887,.519
 321 ,Kurtosis,,.340,1.005
 322 ])
 323
 324 AT_CLEANUP
 325
 326 dnl Test the PERCENTILES subcommand of the EXAMINE command.
 327 dnl In particular test that it behaves properly when there are only
 328 dnl a few cases.
 329 AT_SETUP([EXAMINE -- percentiles])
 330 AT_KEYWORDS([categorical categoricals])
 331 AT_DATA([examine.sps], [dnl
 332 DATA LIST LIST /X *.
 333 BEGIN DATA.
 334 2.00
 335 8.00
 336 5.00
 337 END DATA.
 338
 339 EXAMINE /x
 340         /PERCENTILES=HAVERAGE.
 341
 342 EXAMINE /x
 343         /PERCENTILES=WAVERAGE.
 344
 345 EXAMINE /x
 346         /PERCENTILES=ROUND.
 347
 348 EXAMINE /x
 349         /PERCENTILES=EMPIRICAL.
 350
 351 EXAMINE /x
 352         /PERCENTILES=AEMPIRICAL.
 353 ])
 354 AT_CHECK([pspp -o pspp.csv -o pspp.txt examine.sps])
 355 AT_CHECK([cat pspp.csv], [0], [dnl
 356 Table: Reading free-form data from INLINE.
 357 Variable,Format
 358 X,F8.0
 359
 360 Table: Case Processing Summary
 361 ,Cases,,,,,
 362 ,Valid,,Missing,,Total,
 363 ,N,Percent,N,Percent,N,Percent
 364 X,3,100.0%,0,.0%,3,100.0%
 365
 366 Table: Percentiles
 367 ,,Percentiles,,,,,,
 368 ,,5,10,25,50,75,90,95
 369 X,Weighted Average,.40,.80,2.00,5.00,8.00,8.00,8.00
 370 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 371
 372 Table: Case Processing Summary
 373 ,Cases,,,,,
 374 ,Valid,,Missing,,Total,
 375 ,N,Percent,N,Percent,N,Percent
 376 X,3,100.0%,0,.0%,3,100.0%
 377
 378 Table: Percentiles
 379 ,,Percentiles,,,,,,
 380 ,,5,10,25,50,75,90,95
 381 X,Weighted Average,.30,.60,1.50,3.50,5.75,7.10,7.55
 382 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 383
 384 Table: Case Processing Summary
 385 ,Cases,,,,,
 386 ,Valid,,Missing,,Total,
 387 ,N,Percent,N,Percent,N,Percent
 388 X,3,100.0%,0,.0%,3,100.0%
 389
 390 Table: Percentiles
 391 ,,Percentiles,,,,,,
 392 ,,5,10,25,50,75,90,95
 393 X,Weighted Average,.00,.00,2.00,5.00,5.00,8.00,8.00
 394 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 395
 396 Table: Case Processing Summary
 397 ,Cases,,,,,
 398 ,Valid,,Missing,,Total,
 399 ,N,Percent,N,Percent,N,Percent
 400 X,3,100.0%,0,.0%,3,100.0%
 401
 402 Table: Percentiles
 403 ,,Percentiles,,,,,,
 404 ,,5,10,25,50,75,90,95
 405 X,Weighted Average,2.00,2.00,2.00,5.00,8.00,8.00,8.00
 406 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 407
 408 Table: Case Processing Summary
 409 ,Cases,,,,,
 410 ,Valid,,Missing,,Total,
 411 ,N,Percent,N,Percent,N,Percent
 412 X,3,100.0%,0,.0%,3,100.0%
 413
 414 Table: Percentiles
 415 ,,Percentiles,,,,,,
 416 ,,5,10,25,50,75,90,95
 417 X,Weighted Average,2.00,2.00,2.00,5.00,8.00,8.00,8.00
 418 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 419 ])
 420 AT_CLEANUP
 421
 422 AT_SETUP([EXAMINE -- missing values])
 423 AT_KEYWORDS([categorical categoricals])
 424 AT_DATA([examine.sps], [dnl
 425 DATA LIST LIST /x * y *.
 426 BEGIN DATA.
 427 1   1
 428 2   1
 429 3   1
 430 4   1
 431 5   2
 432 6   2
 433 .   2
 434 END DATA
 435
 436 EXAMINE /x by y
 437         /MISSING = PAIRWISE
 438         .
 439 ])
 440 AT_CHECK([pspp -o pspp.csv examine.sps])
 441 AT_CHECK([cat pspp.csv], [0], [dnl
 442 Table: Reading free-form data from INLINE.
 443 Variable,Format
 444 x,F8.0
 445 y,F8.0
 446
 447 Table: Case Processing Summary
 448 ,Cases,,,,,
 449 ,Valid,,Missing,,Total,
 450 ,N,Percent,N,Percent,N,Percent
 451 x,6,85.7%,1,14.3%,7,100.0%
 452
 453 Table: Case Processing Summary
 454 ,y,Cases,,,,,
 455 ,,Valid,,Missing,,Total,
 456 ,,N,Percent,N,Percent,N,Percent
 457 x,1.00,4,100.0%,0,.0%,4,100.0%
 458 ,2.00,2,66.7%,1,33.3%,3,100.0%
 459 ])
 460 AT_CLEANUP
 461
 462
 463 AT_SETUP([EXAMINE -- user missing values])
 464 AT_KEYWORDS([categorical categoricals])
 465 AT_DATA([examine-m.sps], [dnl
 466 DATA LIST notable LIST /x * y *.
 467 BEGIN DATA.
 468 1                   2
 469 9999999999          2
 470 9999999999          99
 471 END DATA.
 472
 473 MISSING VALUES x (9999999999).
 474 MISSING VALUES y (99).
 475
 476 EXAMINE
 477         /VARIABLES= x y
 478         /MISSING=PAIRWISE.
 479 ])
 480 AT_CHECK([pspp -O format=csv examine-m.sps], [0], [dnl
 481 Table: Case Processing Summary
 482 ,Cases,,,,,
 483 ,Valid,,Missing,,Total,
 484 ,N,Percent,N,Percent,N,Percent
 485 x,1,33.3%,2,66.7%,3,100.0%
 486 y,2,66.7%,1,33.3%,3,100.0%
 487 ])
 488 AT_CLEANUP
 489
 490 AT_SETUP([EXAMINE -- missing values and percentiles])
 491 AT_KEYWORDS([categorical categoricals])
 492 AT_DATA([examine.sps], [dnl
 493 DATA LIST LIST /X *.
 494 BEGIN DATA.
 495 99
 496 99
 497 5.00
 498 END DATA.
 499
 500 MISSING VALUE X (99).
 501
 502 EXAMINE /x
 503         /PERCENTILES=HAVERAGE.
 504 ])
 505 AT_CHECK([pspp -o pspp.csv examine.sps])
 506 dnl Ignore output -- this is just a no-crash check.
 507 AT_CLEANUP
 508
 509 dnl Tests the trimmed mean calculation in the case
 510 dnl where the data is weighted towards the centre.
 511 AT_SETUP([EXAMINE -- trimmed mean])
 512 AT_KEYWORDS([categorical categoricals])
 513 AT_DATA([examine.sps], [dnl
 514 DATA LIST LIST /X * C *.
 515 BEGIN DATA.
 516 1 1
 517 2 49
 518 3 2
 519 END DATA.
 520
 521 WEIGHT BY c.
 522
 523 EXAMINE
 524         x
 525         /STATISTICS=DESCRIPTIVES
 526         .
 527 ])
 528 AT_CHECK([pspp -o pspp.csv examine.sps])
 529 AT_CHECK([cat pspp.csv], [0], [dnl
 530 Table: Reading free-form data from INLINE.
 531 Variable,Format
 532 X,F8.0
 533 C,F8.0
 534
 535 Table: Case Processing Summary
 536 ,Cases,,,,,
 537 ,Valid,,Missing,,Total,
 538 ,N,Percent,N,Percent,N,Percent
 539 X,52.00,100.0%,.00,.0%,52.00,100.0%
 540
 541 Table: Descriptives
 542 ,,,Statistic,Std. Error
 543 X,Mean,,2.02,.03
 544 ,95% Confidence Interval for Mean,Lower Bound,1.95,
 545 ,,Upper Bound,2.09,
 546 ,5% Trimmed Mean,,2.00,
 547 ,Median,,2.00,
 548 ,Variance,,.06,
 549 ,Std. Deviation,,.24,
 550 ,Minimum,,1.00,
 551 ,Maximum,,3.00,
 552 ,Range,,2.00,
 553 ,Interquartile Range,,.00,
 554 ,Skewness,,1.19,.33
 555 ,Kurtosis,,15.73,.65
 556 ])
 557 AT_CLEANUP
 558
 559 AT_SETUP([EXAMINE -- crash bug])
 560 AT_KEYWORDS([categorical categoricals])
 561 AT_DATA([examine.sps], [dnl
 562 data list list /a * x * y *.
 563 begin data.
 564 3 1 3
 565 5 1 4
 566 7 2 3
 567 end data.
 568
 569 examine a by x by y
 570         /statistics=DESCRIPTIVES
 571         .
 572 ])
 573 AT_CHECK([pspp -o pspp.csv examine.sps])
 574 dnl Ignore output -- this is just a no-crash check.
 575 AT_CLEANUP
 576
 577 dnl Test that two consecutive EXAMINE commands don't crash PSPP.
 578 AT_SETUP([EXAMINE -- consecutive runs don't crash])
 579 AT_KEYWORDS([categorical categoricals])
 580 AT_DATA([examine.sps], [dnl
 581 data list list /y * z *.
 582 begin data.
 583 6 4
 584 5 3
 585 7 6
 586 end data.
 587
 588 EXAMINE /VARIABLES= z BY y.
 589
 590 EXAMINE /VARIABLES= z.
 591 ])
 592 AT_CHECK([pspp -o pspp.csv examine.sps])
 593 dnl Ignore output -- this is just a no-crash check.
 594 AT_CLEANUP
 595
 596 dnl Test that /DESCRIPTIVES does not crash in presence of missing values.
 597 AT_SETUP([EXAMINE -- missing values don't crash])
 598 AT_KEYWORDS([categorical categoricals])
 599 AT_DATA([examine.sps], [dnl
 600 data list list /x * y *.
 601 begin data.
 602 1 0
 603 2 0
 604 . 0
 605 3 1
 606 4 1
 607 end data.
 608 examine x by y /statistics=descriptives.
 609 ])
 610 AT_CHECK([pspp -o pspp.csv examine.sps])
 611 dnl Ignore output -- this is just a no-crash check.
 612 AT_CLEANUP
 613
 614 dnl Test that having only a single case doesn't crash.
 615 AT_SETUP([EXAMINE -- single case doesn't crash])
 616 AT_KEYWORDS([categorical categoricals])
 617 AT_DATA([examine.sps], [dnl
 618 DATA LIST LIST /quality * .
 619 BEGIN DATA
 620 3
 621 END DATA
 622
 623
 624 EXAMINE
 625         quality
 626         /STATISTICS descriptives
 627         /PLOT = histogram
 628         .
 629 ])
 630 AT_CHECK([pspp -o pspp.csv examine.sps], [0], [ignore])
 631 dnl Ignore output -- this is just a no-crash check.
 632 AT_CLEANUP
 633
 634 dnl Test that all-missing data doesn't crash.
 635 AT_SETUP([EXAMINE -- all-missing data doesn't crash])
 636 AT_KEYWORDS([categorical categoricals])
 637 AT_DATA([examine.sps], [dnl
 638 DATA LIST LIST /x *.
 639 BEGIN DATA.
 640 .
 641 .
 642 .
 643 .
 644 END DATA.
 645
 646 EXAMINE /x
 647         PLOT=HISTOGRAM BOXPLOT NPPLOT SPREADLEVEL(1) ALL
 648         /ID=x
 649         /STATISTICS = DESCRIPTIVES EXTREME (5) ALL
 650         /PERCENTILE=AEMPIRICAL
 651         .
 652 ])
 653 AT_CHECK([pspp -o pspp.csv examine.sps], [0], [ignore])
 654 dnl Ignore output -- this is just a no-crash check.
 655 AT_CLEANUP
 656
 657 dnl Test that big input doesn't crash (bug 11307).
 658 AT_SETUP([EXAMINE -- big input doesn't crash])
 659 AT_KEYWORDS([categorical categoricals slow])
 660 AT_DATA([examine.sps], [dnl
 661 INPUT PROGRAM.
 662         LOOP #I=1 TO 50000.
 663                 COMPUTE X=NORMAL(10).
 664                 END CASE.
 665         END LOOP.
 666         END FILE.
 667 END INPUT PROGRAM.
 668
 669
 670 EXAMINE /x
 671         /STATISTICS=DESCRIPTIVES.
 672 ])
 673 AT_CHECK([pspp -o pspp.csv examine.sps])
 674 dnl Ignore output -- this is just a no-crash check.
 675 AT_CLEANUP
 676
 677 dnl Another test that big input doesn't crash.
 678 dnl The actual bug that this checks for has been lost.
 679 AT_SETUP([EXAMINE -- big input doesn't crash 2])
 680 AT_KEYWORDS([categorical categoricals slow])
 681 AT_DATA([make-big-input.pl],
 682   [for ($i=0; $i<100000; $i++) { print "AB12\n" };
 683    for ($i=0; $i<100000; $i++) { print "AB04\n" };
 684 ])
 685 AT_CHECK([$PERL make-big-input.pl > large.txt])
 686 AT_DATA([examine.sps], [dnl
 687 DATA LIST FILE='large.txt' /S 1-2 (A) X 3 .
 688
 689
 690 AGGREGATE OUTFILE=* /BREAK=X /A=N.
 691
 692
 693 EXAMINE /A BY X.
 694 ])
 695 AT_CHECK([pspp -o pspp.csv examine.sps])
 696 dnl Ignore output -- this is just a no-crash check.
 697 AT_DATA([more-big-input.pl],
 698   [for ($i=0; $i<25000; $i++) { print "AB04\nAB12\n" };
 699 ])
 700 AT_CHECK([$PERL more-big-input.pl >> large.txt])
 701 AT_CHECK([pspp -o pspp.csv examine.sps])
 702 dnl Ignore output -- this is just a no-crash check.
 703 AT_CLEANUP
 704
 705
 706 dnl Test that the ID command works with non-numberic variables
 707 AT_SETUP([EXAMINE -- non-numeric ID])
 708 AT_KEYWORDS([categorical categoricals])
 709
 710 AT_DATA([examine-id.sps], [dnl
 711 data list notable list /x * y (a12).
 712 begin data.
 713 1  one
 714 2  two
 715 3  three
 716 4  four
 717 5  five
 718 6  six
 719 7  seven
 720 8  eight
 721 9  nine
 722 10 ten
 723 11 eleven
 724 12 twelve
 725 30 thirty
 726 300 threehundred
 727 end data.
 728
 729 set small=0.
 730 examine x
 731         /statistics = extreme
 732         /id = y
 733         /plot = boxplot
 734         .
 735 ])
 736
 737 AT_CHECK([pspp -O format=csv examine-id.sps], [0], [dnl
 738 Table: Case Processing Summary
 739 ,Cases,,,,,
 740 ,Valid,,Missing,,Total,
 741 ,N,Percent,N,Percent,N,Percent
 742 x,14,100.0%,0,.0%,14,100.0%
 743
 744 Table: Extreme Values
 745 ,,,y,Value
 746 x,Highest,1,threehundred,300.00
 747 ,,2,thirty,30.00
 748 ,,3,twelve,12.00
 749 ,,4,eleven,11.00
 750 ,,5,ten,10.00
 751 ,Lowest,1,one,1.00
 752 ,,2,two,2.00
 753 ,,3,three,3.00
 754 ,,4,four,4.00
 755 ,,5,five,5.00
 756
 757 Table: Tests of Normality
 758 ,Shapiro-Wilk,,
 759 ,Statistic,df,Sig.
 760 x,.37,14,.00
 761 ])
 762
 763 AT_CLEANUP
 764
 765 dnl Test for a crash which happened on cleanup from a bad input syntax
 766 AT_SETUP([EXAMINE -- Bad Input])
 767 AT_KEYWORDS([categorical categoricals])
 768
 769 AT_DATA([examine-bad.sps], [dnl
 770 data list list /h * g *.
 771 begin data.
 772 1 1
 773 2 1
 774 3 1
 775 4 1
 776 5 2
 777 6 2
 778 7 2
 779 8 2
 780 9 2
 781 end data.
 782
 783 EXAMINE
 784         /VARIABLES= h
 785         BY  g
 786         /STATISTICS = DESCRIPTIVES EXTREME
 787         /PLOT = lkajsdas
 788         .
 789 ])
 790
 791 AT_CHECK([pspp -o pspp.csv examine-bad.sps], [1], [ignore])
 792
 793 AT_CLEANUP
 794
 795
 796 dnl Check the MISSING=REPORT option
 797 AT_SETUP([EXAMINE -- MISSING=REPORT])
 798 AT_KEYWORDS([categorical categoricals])
 799
 800 AT_DATA([examine-report.sps], [dnl
 801 set format = F22.0.
 802 data list list /x * g *.
 803 begin data.
 804 1   1
 805 2   1
 806 3   1
 807 4   1
 808 5   1
 809 6   1
 810 7   1
 811 8   1
 812 9   1
 813 10   2
 814 20   2
 815 30   2
 816 40   2
 817 50   2
 818 60   2
 819 70   2
 820 80   2
 821 90   2
 822 101   9
 823 201   9
 824 301   9
 825 401   9
 826 501   99
 827 601   99
 828 701   99
 829 801   99
 830 901   99
 831 1001  .
 832 2002  .
 833 3003  .
 834 4004  .
 835 end data.
 836
 837 MISSING VALUES g (9, 99, 999).
 838
 839 EXAMINE
 840         /VARIABLES = x
 841         BY  g
 842         /STATISTICS = EXTREME
 843         /NOTOTAL
 844         /MISSING = REPORT.
 845 ])
 846
 847
 848 AT_CHECK([pspp -o pspp.csv -o pspp.txt examine-report.sps])
 849 AT_CHECK([cat pspp.csv], [0],
 850   [[Table: Reading free-form data from INLINE.
 851 Variable,Format
 852 x,F8.0
 853 g,F8.0
 854
 855 Table: Case Processing Summary
 856 ,g,Cases,,,,,
 857 ,,Valid,,Missing,,Total,
 858 ,,N,Percent,N,Percent,N,Percent
 859 x,.,4,100.0%,0,.0%,4,100.0%
 860 ,1,9,100.0%,0,.0%,9,100.0%
 861 ,2,9,100.0%,0,.0%,9,100.0%
 862 ,9[a],4,100.0%,0,.0%,4,100.0%
 863 ,99[a],5,100.0%,0,.0%,5,100.0%
 864 Footnote: a. User-missing value.
 865
 866 Table: Extreme Values
 867 ,g,,,Case Number,Value
 868 x,.,Highest,1,31,4004
 869 ,,,2,30,3003
 870 ,,,3,29,2002
 871 ,,,4,28,1001
 872 ,,,5,0,0
 873 ,,Lowest,1,28,1001
 874 ,,,2,29,2002
 875 ,,,3,30,3003
 876 ,,,4,31,4004
 877 ,,,5,31,4004
 878 ,1,Highest,1,9,9
 879 ,,,2,8,8
 880 ,,,3,7,7
 881 ,,,4,6,6
 882 ,,,5,5,5
 883 ,,Lowest,1,1,1
 884 ,,,2,2,2
 885 ,,,3,3,3
 886 ,,,4,4,4
 887 ,,,5,5,5
 888 ,2,Highest,1,18,90
 889 ,,,2,17,80
 890 ,,,3,16,70
 891 ,,,4,15,60
 892 ,,,5,14,50
 893 ,,Lowest,1,10,10
 894 ,,,2,11,20
 895 ,,,3,12,30
 896 ,,,4,13,40
 897 ,,,5,14,50
 898 ,9[a],Highest,1,22,401
 899 ,,,2,21,301
 900 ,,,3,20,201
 901 ,,,4,19,101
 902 ,,,5,0,0
 903 ,,Lowest,1,19,101
 904 ,,,2,20,201
 905 ,,,3,21,301
 906 ,,,4,22,401
 907 ,,,5,22,401
 908 ,99[a],Highest,1,27,901
 909 ,,,2,26,801
 910 ,,,3,25,701
 911 ,,,4,24,601
 912 ,,,5,23,501
 913 ,,Lowest,1,23,501
 914 ,,,2,24,601
 915 ,,,3,25,701
 916 ,,,4,26,801
 917 ,,,5,27,901
 918 Footnote: a. User-missing value.
 919 ]])
 920
 921 AT_CLEANUP
 922
 923
 924 dnl Run a test of the basic STATISTICS using a "real"
 925 dnl dataset and comparing with "real" results kindly
 926 dnl provided by Olaf Nöhring
 927 AT_SETUP([EXAMINE -- sample unweighted])
 928 AT_KEYWORDS([categorical categoricals])
 929
 930 AT_DATA([sample.sps], [dnl
 931 set format = F22.4.
 932 DATA LIST notable LIST /X *
 933 BEGIN DATA.
 934 461.19000000
 935 466.38000000
 936 479.46000000
 937 480.10000000
 938 483.43000000
 939 488.30000000
 940 489.00000000
 941 491.62000000
 942 505.62000000
 943 511.30000000
 944 521.53000000
 945 526.70000000
 946 528.25000000
 947 538.70000000
 948 540.22000000
 949 540.58000000
 950 546.10000000
 951 548.17000000
 952 553.99000000
 953 566.21000000
 954 575.90000000
 955 584.38000000
 956 593.40000000
 957 357.05000000
 958 359.73000000
 959 360.48000000
 960 373.98000000
 961 374.13000000
 962 381.45000000
 963 383.72000000
 964 390.00000000
 965 400.34000000
 966 415.32000000
 967 415.91000000
 968 418.30000000
 969 421.03000000
 970 422.43000000
 971 426.93000000
 972 433.25000000
 973 436.89000000
 974 445.33000000
 975 446.33000000
 976 446.55000000
 977 456.44000000
 978 689.49000000
 979 691.92000000
 980 695.00000000
 981 695.36000000
 982 698.21000000
 983 699.46000000
 984 706.61000000
 985 710.69000000
 986 715.82000000
 987 715.82000000
 988 741.39000000
 989 752.27000000
 990 756.73000000
 991 757.74000000
 992 759.57000000
 993 796.07000000
 994 813.78000000
 995 817.25000000
 996 825.48000000
 997 831.28000000
 998 849.24000000
 999 890.00000000
1000 894.78000000
1001 935.65000000
1002 935.90000000
1003 945.90000000
1004 1012.8600000
1005 1022.6000000
1006 1061.8100000
1007 1063.5000000
1008 1077.2300000
1009 1151.6300000
1010 1355.2800000
1011 598.88000000
1012 606.91000000
1013 621.60000000
1014 624.80000000
1015 636.13000000
1016 637.38000000
1017 640.32000000
1018 649.35000000
1019 656.51000000
1020 662.55000000
1021 664.69000000
1022 106.22000000
1023 132.24000000
1024 174.76000000
1025 204.85000000
1026 264.93000000
1027 264.99000000
1028 269.84000000
1029 325.12000000
1030 331.67000000
1031 337.26000000
1032 347.68000000
1033 354.91000000
1034 END DATA.
1035
1036 EXAMINE
1037         x
1038         /STATISTICS=DESCRIPTIVES
1039         .
1040 ])
1041
1042 AT_CHECK([pspp -O format=csv sample.sps], [0], [dnl
1043 Table: Case Processing Summary
1044 ,Cases,,,,,
1045 ,Valid,,Missing,,Total,
1046 ,N,Percent,N,Percent,N,Percent
1047 X,100,100.0%,0,.0%,100,100.0%
1048
1049 Table: Descriptives
1050 ,,,Statistic,Std. Error
1051 X,Mean,,587.6603,23.2665
1052 ,95% Confidence Interval for Mean,Lower Bound,541.4946,
1053 ,,Upper Bound,633.8260,
1054 ,5% Trimmed Mean,,579.7064,
1055 ,Median,,547.1350,
1056 ,Variance,,54132.8466,
1057 ,Std. Deviation,,232.6647,
1058 ,Minimum,,106.2200,
1059 ,Maximum,,1355.2800,
1060 ,Range,,1249.0600,
1061 ,Interquartile Range,,293.1575,
1062 ,Skewness,,.6331,.2414
1063 ,Kurtosis,,.5300,.4783
1064 ])
1065
1066 AT_CLEANUP
1067
1068
1069
1070 dnl Test for a crash which happened on bad input syntax
1071 AT_SETUP([EXAMINE -- Empty Parentheses])
1072 AT_KEYWORDS([categorical categoricals])
1073
1074 AT_DATA([examine-empty-parens.sps], [dnl
1075 DATA LIST notable LIST /X *
1076 BEGIN DATA.
1077 2
1078 3
1079 END DATA.
1080
1081
1082 EXAMINE
1083         x
1084         /PLOT = SPREADLEVEL()
1085         .
1086 ])
1087
1088 AT_CHECK([pspp -o pspp.csv examine-empty-parens.sps], [1], [ignore])
1089
1090 AT_CLEANUP
1091
1092
1093
1094
1095 dnl Test for another crash which happened on bad input syntax
1096 AT_SETUP([EXAMINE -- Bad variable])
1097 AT_KEYWORDS([categorical categoricals])
1098
1099 AT_DATA([examine-bad-variable.sps], [dnl
1100 data list list /h * g *.
1101 begin data.
1102 3 1
1103 4 1
1104 5 2
1105 end data.
1106
1107 EXAMINE
1108         /VARIABLES/ h
1109         BY  g
1110         .
1111 ])
1112
1113 AT_CHECK([pspp -o pspp.csv examine-bad-variable.sps], [1], [ignore])
1114
1115 AT_CLEANUP
1116
1117
1118
1119 dnl Test for yet another crash. This time for extremes vs. missing weight values.\0
1120 AT_SETUP([EXAMINE -- Extremes vs. Missing Weights])
1121 AT_KEYWORDS([categorical categoricals])
1122
1123 AT_DATA([examine-missing-weights.sps], [dnl
1124 data list notable list /h * g *.
1125 begin data.
1126 3 1
1127 4 .
1128 5 1
1129 2 1
1130 end data.
1131
1132 WEIGHT BY g.
1133
1134 EXAMINE h
1135         /STATISTICS extreme(3)
1136         .
1137 ])
1138
1139 AT_CHECK([pspp -O format=csv  examine-missing-weights.sps], [0], [dnl
1140 "examine-missing-weights.sps:13: warning: EXAMINE: At least one case in the data file had a weight value that was user-missing, system-missing, zero, or negative.  These case(s) were ignored."
1141
1142 Table: Case Processing Summary
1143 ,Cases,,,,,
1144 ,Valid,,Missing,,Total,
1145 ,N,Percent,N,Percent,N,Percent
1146 h,3.00,100.0%,.00,.0%,3.00,100.0%
1147
1148 Table: Extreme Values
1149 ,,,Case Number,Value
1150 h,Highest,1,3,5.00
1151 ,,2,2,4.00
1152 ,,3,1,3.00
1153 ,Lowest,1,4,2.00
1154 ,,2,1,3.00
1155 ,,3,2,4.00
1156 ])
1157
1158 AT_CLEANUP
1159
1160 dnl This is an example from doc/tutorial.texi
1161 dnl So if the results of this have to be changed in any way,
1162 dnl make sure to update that file.
1163 AT_SETUP([EXAMINE tutorial example 1])
1164 cp $top_srcdir/examples/repairs.sav .
1165 AT_DATA([repairs.sps], [dnl
1166 GET FILE='repairs.sav'.
1167 EXAMINE mtbf /STATISTICS=DESCRIPTIVES.
1168 COMPUTE mtbf_ln = LN (mtbf).
1169 EXAMINE mtbf_ln /STATISTICS=DESCRIPTIVES.
1170 ])
1171
1172 AT_CHECK([pspp -O format=csv repairs.sps], [0], [dnl
1173 Table: Case Processing Summary
1174 ,Cases,,,,,
1175 ,Valid,,Missing,,Total,
1176 ,N,Percent,N,Percent,N,Percent
1177 Mean time between failures (months) ,30,100.0%,0,.0%,30,100.0%
1178
1179 Table: Descriptives
1180 ,,,Statistic,Std. Error
1181 Mean time between failures (months) ,Mean,,8.78,1.10
1182 ,95% Confidence Interval for Mean,Lower Bound,6.53,
1183 ,,Upper Bound,11.04,
1184 ,5% Trimmed Mean,,8.20,
1185 ,Median,,8.29,
1186 ,Variance,,36.34,
1187 ,Std. Deviation,,6.03,
1188 ,Minimum,,1.63,
1189 ,Maximum,,26.47,
1190 ,Range,,24.84,
1191 ,Interquartile Range,,6.03,
1192 ,Skewness,,1.65,.43
1193 ,Kurtosis,,3.41,.83
1194
1195 Table: Case Processing Summary
1196 ,Cases,,,,,
1197 ,Valid,,Missing,,Total,
1198 ,N,Percent,N,Percent,N,Percent
1199 mtbf_ln,30,100.0%,0,.0%,30,100.0%
1200
1201 Table: Descriptives
1202 ,,,Statistic,Std. Error
1203 mtbf_ln,Mean,,1.95,.13
1204 ,95% Confidence Interval for Mean,Lower Bound,1.69,
1205 ,,Upper Bound,2.22,
1206 ,5% Trimmed Mean,,1.96,
1207 ,Median,,2.11,
1208 ,Variance,,.49,
1209 ,Std. Deviation,,.70,
1210 ,Minimum,,.49,
1211 ,Maximum,,3.28,
1212 ,Range,,2.79,
1213 ,Interquartile Range,,.88,
1214 ,Skewness,,-.37,.43
1215 ,Kurtosis,,.01,.83
1216 ])
1217
1218 AT_CLEANUP
1219
1220 dnl This is an example from doc/tutorial.texi
1221 dnl So if the results of this have to be changed in any way,
1222 dnl make sure to update that file.
1223 AT_SETUP([EXAMINE tutorial example 2])
1224 cp $top_srcdir/examples/physiology.sav .
1225 AT_DATA([examine.sps], [dnl
1226 GET FILE='physiology.sav'.
1227 EXAMINE height, weight /STATISTICS=EXTREME(3).
1228 ])
1229 AT_CHECK([pspp -o pspp.csv -o pspp.txt examine.sps])
1230 AT_CHECK([cat pspp.csv], [0], [dnl
1231 Table: Case Processing Summary
1232 ,Cases,,,,,
1233 ,Valid,,Missing,,Total,
1234 ,N,Percent,N,Percent,N,Percent
1235 Height in millimeters   ,40,100.0%,0,.0%,40,100.0%
1236 Weight in kilograms ,40,100.0%,0,.0%,40,100.0%
1237
1238 Table: Extreme Values
1239 ,,,Case Number,Value
1240 Height in millimeters   ,Highest,1,14,1903
1241 ,,2,15,1884
1242 ,,3,12,1802
1243 ,Lowest,1,30,179
1244 ,,2,31,1598
1245 ,,3,28,1601
1246 Weight in kilograms ,Highest,1,13,92.1
1247 ,,2,5,92.1
1248 ,,3,17,91.7
1249 ,Lowest,1,38,-55.6
1250 ,,2,39,54.5
1251 ,,3,33,55.4
1252 ])
1253 AT_CLEANUP
1254
1255
1256
1257 AT_SETUP([EXAMINE -- Crash on unrepresentable graphs])
1258 AT_DATA([examine.sps], [dnl
1259 data list notable list /x * g *.
1260 begin data.
1261 96 1
1262 end data.
1263
1264 examine x  by g
1265         /nototal
1266         /plot = all.
1267 ])
1268 dnl This bug only manifested itself on cairo based drivers.
1269 AT_CHECK([pspp -O format=pdf examine.sps], [0], [ignore], [ignore])
1270 AT_CLEANUP
1271
1272
1273 dnl This example comes from the web site:
1274 dnl  https://www.spsstests.com/2018/11/shapiro-wilk-normality-test-spss.html
1275 AT_SETUP([EXAMINE -- shapiro-wilk 1])
1276 AT_KEYWORDS([shapiro wilk])
1277 AT_DATA([shapiro-wilk.sps], [dnl
1278 data list notable list /x * g *.
1279 begin data.
1280 96 1
1281 98 1
1282 95 1
1283 89 1
1284 90 1
1285 92 1
1286 94 1
1287 93 1
1288 97 1
1289 100 1
1290 99 2
1291 96 2
1292 80 2
1293 89 2
1294 91 2
1295 92 2
1296 93 2
1297 94 2
1298 99 2
1299 80 2
1300 end data.
1301
1302 set format F22.3.
1303
1304 examine x  by g
1305         /nototal
1306         /plot = all.
1307 ])
1308
1309 AT_CHECK([pspp -O format=csv shapiro-wilk.sps], [0],[dnl
1310 Table: Case Processing Summary
1311 ,g,Cases,,,,,
1312 ,,Valid,,Missing,,Total,
1313 ,,N,Percent,N,Percent,N,Percent
1314 x,1.00,10,100.0%,0,.0%,10,100.0%
1315 ,2.00,10,100.0%,0,.0%,10,100.0%
1316
1317 Table: Tests of Normality
1318 ,g,Shapiro-Wilk,,
1319 ,,Statistic,df,Sig.
1320 x,1.00,.984,10,.983
1321 ,2.00,.882,10,.136
1322 ])
1323
1324 AT_CLEANUP
1325
1326
1327 dnl This example comes from the web site:
1328 dnl  http://www.real-statistics.com/tests-normality-and-symmetry/statistical-tests-normality-symmetry/shapiro-wilk-expanded-test/
1329 dnl It uses a dataset larger than 11 samples. Hence the alternative method for
1330 dnl signficance is used.
1331 AT_SETUP([EXAMINE -- shapiro-wilk 2])
1332 AT_KEYWORDS([shapiro wilk])
1333 AT_DATA([shapiro-wilk2.sps], [dnl
1334 data list notable list /x *.
1335 begin data.
1336 65
1337 61
1338 63
1339 86
1340 70
1341 55
1342 74
1343 35
1344 72
1345 68
1346 45
1347 58
1348 end data.
1349
1350 set format F22.3.
1351
1352 examine x
1353         /plot = boxplot.
1354 ])
1355
1356 AT_CHECK([pspp -O format=csv shapiro-wilk2.sps], [0],[dnl
1357 Table: Case Processing Summary
1358 ,Cases,,,,,
1359 ,Valid,,Missing,,Total,
1360 ,N,Percent,N,Percent,N,Percent
1361 x,12,100.0%,0,.0%,12,100.0%
1362
1363 Table: Tests of Normality
1364 ,Shapiro-Wilk,,
1365 ,Statistic,df,Sig.
1366 x,.971,12,.922
1367 ])
1368
1369 AT_CLEANUP