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

   1 dnl PSPP - a program for statistical analysis.
   2 dnl Copyright (C) 2017 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 AT_BANNER([EXAMINE])
  17
  18 AT_SETUP([EXAMINE])
  19 AT_DATA([examine.sps], [
  20 DATA LIST LIST /QUALITY * W * BRAND * .
  21 BEGIN DATA
  22 3  1  1
  23 2  2  1
  24 1  2  1
  25 1  1  1
  26 4  1  1
  27 4  1  1
  28 5  1  2
  29 2  1  2
  30 4  4  2
  31 2  1  2
  32 3  1  2
  33 7  1  3
  34 4  2  3
  35 5  3  3
  36 3  1  3
  37 6  1  3
  38 END DATA
  39
  40 WEIGHT BY w.
  41
  42 VARIABLE LABELS brand   'Manufacturer'.
  43 VARIABLE LABELS quality 'Breaking Strain'.
  44
  45 VALUE LABELS /brand 1 'Aspeger' 2 'Bloggs' 3 'Charlies'.
  46
  47 LIST /FORMAT=NUMBERED.
  48
  49 EXAMINE
  50         quality BY brand
  51         /STATISTICS descriptives extreme(3)
  52         .
  53 ])
  54
  55
  56 dnl In the following data, only the extreme values have been checked.
  57 dnl The descriptives have been blindly pasted.
  58 AT_CHECK([pspp -O format=csv examine.sps], [0], [dnl
  59 Table: Reading free-form data from INLINE.
  60 Variable,Format
  61 QUALITY,F8.0
  62 W,F8.0
  63 BRAND,F8.0
  64
  65 Table: Data List
  66 Case Number,QUALITY,W,BRAND
  67 1,3.00,1.00,1.00
  68 2,2.00,2.00,1.00
  69 3,1.00,2.00,1.00
  70 4,1.00,1.00,1.00
  71 5,4.00,1.00,1.00
  72 6,4.00,1.00,1.00
  73 7,5.00,1.00,2.00
  74 8,2.00,1.00,2.00
  75 9,4.00,4.00,2.00
  76 10,2.00,1.00,2.00
  77 11,3.00,1.00,2.00
  78 12,7.00,1.00,3.00
  79 13,4.00,2.00,3.00
  80 14,5.00,3.00,3.00
  81 15,3.00,1.00,3.00
  82 16,6.00,1.00,3.00
  83
  84 Table: Case Processing Summary
  85 ,Cases,,,,,
  86 ,Valid,,Missing,,Total,
  87 ,N,Percent,N,Percent,N,Percent
  88 Breaking Strain,24.00,100%,.00,0%,24.00,100%
  89
  90 Table: Extreme Values
  91 ,,,Case Number,Value
  92 Breaking Strain,Highest,1,12,7.00
  93 ,,2,16,6.00
  94 ,,3,14,5.00
  95 ,Lowest,1,3,1.00
  96 ,,2,4,1.00
  97 ,,3,2,2.00
  98
  99 Table: Descriptives
 100 ,,,Statistic,Std. Error
 101 Breaking Strain,Mean,,3.54,.32
 102 ,95% Confidence Interval for Mean,Lower Bound,2.87,
 103 ,,Upper Bound,4.21,
 104 ,5% Trimmed Mean,,3.50,
 105 ,Median,,4.00,
 106 ,Variance,,2.52,
 107 ,Std. Deviation,,1.59,
 108 ,Minimum,,1.00,
 109 ,Maximum,,7.00,
 110 ,Range,,6.00,
 111 ,Interquartile Range,,2.75,
 112 ,Skewness,,.06,.47
 113 ,Kurtosis,,-.36,.92
 114
 115 Table: Case Processing Summary
 116 ,,Cases,,,,,
 117 ,,Valid,,Missing,,Total,
 118 ,Manufacturer,N,Percent,N,Percent,N,Percent
 119 Breaking Strain,Aspeger,8.00,100%,.00,0%,8.00,100%
 120 ,Bloggs,8.00,100%,.00,0%,8.00,100%
 121 ,Charlies,8.00,100%,.00,0%,8.00,100%
 122
 123 Table: Extreme Values
 124 ,Manufacturer,,,Case Number,Value
 125 Breaking Strain,Aspeger,Highest,1,6,4.00
 126 ,,,2,5,4.00
 127 ,,,3,1,3.00
 128 ,,Lowest,1,3,1.00
 129 ,,,2,4,1.00
 130 ,,,3,2,2.00
 131 ,Bloggs,Highest,1,7,5.00
 132 ,,,2,9,4.00
 133 ,,,3,11,3.00
 134 ,,Lowest,1,8,2.00
 135 ,,,2,10,2.00
 136 ,,,3,11,3.00
 137 ,Charlies,Highest,1,12,7.00
 138 ,,,2,16,6.00
 139 ,,,3,14,5.00
 140 ,,Lowest,1,15,3.00
 141 ,,,2,13,4.00
 142 ,,,3,14,5.00
 143
 144 Table: Descriptives
 145 ,Manufacturer,,,Statistic,Std. Error
 146 Breaking Strain,Aspeger,Mean,,2.25,.45
 147 ,,95% Confidence Interval for Mean,Lower Bound,1.18,
 148 ,,,Upper Bound,3.32,
 149 ,,5% Trimmed Mean,,2.22,
 150 ,,Median,,2.00,
 151 ,,Variance,,1.64,
 152 ,,Std. Deviation,,1.28,
 153 ,,Minimum,,1.00,
 154 ,,Maximum,,4.00,
 155 ,,Range,,3.00,
 156 ,,Interquartile Range,,2.75,
 157 ,,Skewness,,.47,.75
 158 ,,Kurtosis,,-1.55,1.48
 159 ,Bloggs,Mean,,3.50,.38
 160 ,,95% Confidence Interval for Mean,Lower Bound,2.61,
 161 ,,,Upper Bound,4.39,
 162 ,,5% Trimmed Mean,,3.50,
 163 ,,Median,,4.00,
 164 ,,Variance,,1.14,
 165 ,,Std. Deviation,,1.07,
 166 ,,Minimum,,2.00,
 167 ,,Maximum,,5.00,
 168 ,,Range,,3.00,
 169 ,,Interquartile Range,,1.75,
 170 ,,Skewness,,-.47,.75
 171 ,,Kurtosis,,-.83,1.48
 172 ,Charlies,Mean,,4.88,.44
 173 ,,95% Confidence Interval for Mean,Lower Bound,3.83,
 174 ,,,Upper Bound,5.92,
 175 ,,5% Trimmed Mean,,4.86,
 176 ,,Median,,5.00,
 177 ,,Variance,,1.55,
 178 ,,Std. Deviation,,1.25,
 179 ,,Minimum,,3.00,
 180 ,,Maximum,,7.00,
 181 ,,Range,,4.00,
 182 ,,Interquartile Range,,1.75,
 183 ,,Skewness,,.30,.75
 184 ,,Kurtosis,,.15,1.48
 185 ])
 186
 187 AT_CLEANUP
 188
 189 AT_SETUP([EXAMINE -- extremes])
 190 AT_DATA([examine.sps], [dnl
 191 data list free /V1 W
 192 begin data.
 193 1  1
 194 2  1
 195 3  2
 196 3  1
 197 4  1
 198 5  1
 199 6  1
 200 7  1
 201 8  1
 202 9  1
 203 10 1
 204 11 1
 205 12 1
 206 13 1
 207 14 1
 208 15 1
 209 16 1
 210 17 1
 211 18 2
 212 19 1
 213 20 1
 214 end data.
 215
 216 weight by w.
 217
 218 examine v1
 219  /statistics=extreme(6)
 220  .
 221 ])
 222
 223 AT_CHECK([pspp -O format=csv examine.sps], [0],[dnl
 224 Table: Case Processing Summary
 225 ,Cases,,,,,
 226 ,Valid,,Missing,,Total,
 227 ,N,Percent,N,Percent,N,Percent
 228 V1,23.00,100%,.00,0%,23.00,100%
 229
 230 Table: Extreme Values
 231 ,,,Case Number,Value
 232 V1,Highest,1,21,20.00
 233 ,,2,20,19.00
 234 ,,3,19,18.00
 235 ,,4,18,17.00
 236 ,,5,17,16.00
 237 ,,6,16,15.00
 238 ,Lowest,1,1,1.00
 239 ,,2,2,2.00
 240 ,,3,3,3.00
 241 ,,4,4,3.00
 242 ,,5,5,4.00
 243 ,,6,6,5.00
 244 ])
 245
 246 AT_CLEANUP
 247
 248
 249
 250 AT_SETUP([EXAMINE -- extremes with fractional weights])
 251 AT_DATA([extreme.sps], [dnl
 252 set format=F20.3.
 253 data list notable list /w * x *.
 254 begin data.
 255  0.88  300000
 256  0.86  320000
 257  0.98  480000
 258  0.93  960000
 259  1.35  960000
 260  1.31  960000
 261  0.88  960000
 262  0.88  1080000
 263  0.88  1080000
 264  0.95  1200000
 265  1.47  1200000
 266  0.93  1200000
 267  0.98  1320000
 268  1.31  1380000
 269  0.93  1440000
 270  0.88  1560000
 271  1.56  1560000
 272  1.47  1560000
 273 end data.
 274
 275 weight by w.
 276
 277
 278 EXAMINE
 279         x
 280         /STATISTICS = DESCRIPTIVES EXTREME (5)
 281         .
 282 ])
 283
 284 AT_CHECK([pspp -O format=csv  extreme.sps], [0], [dnl
 285 Table: Case Processing Summary
 286 ,Cases,,,,,
 287 ,Valid,,Missing,,Total,
 288 ,N,Percent,N,Percent,N,Percent
 289 x,19.430,100%,.000,0%,19.430,100%
 290
 291 Table: Extreme Values
 292 ,,,Case Number,Value
 293 x,Highest,1,18,1560000.000
 294 ,,2,17,1560000.000
 295 ,,3,16,1560000.000
 296 ,,4,15,1440000.000
 297 ,,5,14,1380000.000
 298 ,Lowest,1,1,300000.000
 299 ,,2,2,320000.000
 300 ,,3,3,480000.000
 301 ,,4,4,960000.000
 302 ,,5,5,960000.000
 303
 304 Table: Descriptives
 305 ,,,Statistic,Std. Error
 306 x,Mean,,1120010.293,86222.178
 307 ,95% Confidence Interval for Mean,Lower Bound,939166.693,
 308 ,,Upper Bound,1300853.894,
 309 ,5% Trimmed Mean,,1141017.899,
 310 ,Median,,1200000.000,
 311 ,Variance,,144447748124.869,
 312 ,Std. Deviation,,380062.821,
 313 ,Minimum,,300000.000,
 314 ,Maximum,,1560000.000,
 315 ,Range,,1260000.000,
 316 ,Interquartile Range,,467258.065,
 317 ,Skewness,,-.887,.519
 318 ,Kurtosis,,.340,1.005
 319 ])
 320
 321 AT_CLEANUP
 322
 323 dnl Test the PERCENTILES subcommand of the EXAMINE command.
 324 dnl In particular test that it behaves properly when there are only
 325 dnl a few cases.
 326 AT_SETUP([EXAMINE -- percentiles])
 327 AT_DATA([examine.sps], [dnl
 328 DATA LIST LIST /X *.
 329 BEGIN DATA.
 330 2.00
 331 8.00
 332 5.00
 333 END DATA.
 334
 335 EXAMINE /x
 336         /PERCENTILES=HAVERAGE.
 337
 338 EXAMINE /x
 339         /PERCENTILES=WAVERAGE.
 340
 341 EXAMINE /x
 342         /PERCENTILES=ROUND.
 343
 344 EXAMINE /x
 345         /PERCENTILES=EMPIRICAL.
 346
 347 EXAMINE /x
 348         /PERCENTILES=AEMPIRICAL.
 349 ])
 350 AT_CHECK([pspp -o pspp.csv examine.sps])
 351 AT_CHECK([cat pspp.csv], [0], [dnl
 352 Table: Reading free-form data from INLINE.
 353 Variable,Format
 354 X,F8.0
 355
 356 Table: Case Processing Summary
 357 ,Cases,,,,,
 358 ,Valid,,Missing,,Total,
 359 ,N,Percent,N,Percent,N,Percent
 360 X,3,100%,0,0%,3,100%
 361
 362 Table: Percentiles
 363 ,,Percentiles,,,,,,
 364 ,,5,10,25,50,75,90,95
 365 X,HAverage,.40,.80,2.00,5.00,8.00,8.00,8.00
 366 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 367
 368 Table: Case Processing Summary
 369 ,Cases,,,,,
 370 ,Valid,,Missing,,Total,
 371 ,N,Percent,N,Percent,N,Percent
 372 X,3,100%,0,0%,3,100%
 373
 374 Table: Percentiles
 375 ,,Percentiles,,,,,,
 376 ,,5,10,25,50,75,90,95
 377 X,Weighted Average,.30,.60,1.50,3.50,5.75,7.10,7.55
 378 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 379
 380 Table: Case Processing Summary
 381 ,Cases,,,,,
 382 ,Valid,,Missing,,Total,
 383 ,N,Percent,N,Percent,N,Percent
 384 X,3,100%,0,0%,3,100%
 385
 386 Table: Percentiles
 387 ,,Percentiles,,,,,,
 388 ,,5,10,25,50,75,90,95
 389 X,Rounded,.00,.00,2.00,5.00,5.00,8.00,8.00
 390 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 391
 392 Table: Case Processing Summary
 393 ,Cases,,,,,
 394 ,Valid,,Missing,,Total,
 395 ,N,Percent,N,Percent,N,Percent
 396 X,3,100%,0,0%,3,100%
 397
 398 Table: Percentiles
 399 ,,Percentiles,,,,,,
 400 ,,5,10,25,50,75,90,95
 401 X,Empirical,2.00,2.00,2.00,5.00,8.00,8.00,8.00
 402 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 403
 404 Table: Case Processing Summary
 405 ,Cases,,,,,
 406 ,Valid,,Missing,,Total,
 407 ,N,Percent,N,Percent,N,Percent
 408 X,3,100%,0,0%,3,100%
 409
 410 Table: Percentiles
 411 ,,Percentiles,,,,,,
 412 ,,5,10,25,50,75,90,95
 413 X,Empirical with averaging,2.00,2.00,2.00,5.00,8.00,8.00,8.00
 414 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 415 ])
 416 AT_CLEANUP
 417
 418 AT_SETUP([EXAMINE -- missing values])
 419 AT_DATA([examine.sps], [dnl
 420 DATA LIST LIST /x * y *.
 421 BEGIN DATA.
 422 1   1
 423 2   1
 424 3   1
 425 4   1
 426 5   2
 427 6   2
 428 .   2
 429 END DATA
 430
 431 EXAMINE /x by y
 432         /MISSING = PAIRWISE
 433         .
 434 ])
 435 AT_CHECK([pspp -o pspp.csv examine.sps])
 436 AT_CHECK([cat pspp.csv], [0], [dnl
 437 Table: Reading free-form data from INLINE.
 438 Variable,Format
 439 x,F8.0
 440 y,F8.0
 441
 442 Table: Case Processing Summary
 443 ,Cases,,,,,
 444 ,Valid,,Missing,,Total,
 445 ,N,Percent,N,Percent,N,Percent
 446 x,6,85.7143%,1,14.2857%,7,100%
 447
 448 Table: Case Processing Summary
 449 ,,Cases,,,,,
 450 ,,Valid,,Missing,,Total,
 451 ,y,N,Percent,N,Percent,N,Percent
 452 x,1.00,4,100%,0,0%,4,100%
 453 ,2.00,2,66.6667%,1,33.3333%,3,100%
 454 ])
 455 AT_CLEANUP
 456
 457
 458 AT_SETUP([EXAMINE -- user missing values])
 459 AT_DATA([examine-m.sps], [dnl
 460 DATA LIST notable LIST /x * y *.
 461 BEGIN DATA.
 462 1                   2
 463 9999999999          2
 464 9999999999          99
 465 END DATA.
 466
 467 MISSING VALUES x (9999999999).
 468 MISSING VALUES y (99).
 469
 470 EXAMINE
 471         /VARIABLES= x y
 472         /MISSING=PAIRWISE.
 473 ])
 474 AT_CHECK([pspp -O format=csv examine-m.sps], [0], [dnl
 475 Table: Case Processing Summary
 476 ,Cases,,,,,
 477 ,Valid,,Missing,,Total,
 478 ,N,Percent,N,Percent,N,Percent
 479 x,1,33.3333%,2,66.6667%,3,100%
 480 y,2,66.6667%,1,33.3333%,3,100%
 481 ])
 482 AT_CLEANUP
 483
 484 AT_SETUP([EXAMINE -- missing values and percentiles])
 485 AT_DATA([examine.sps], [dnl
 486 DATA LIST LIST /X *.
 487 BEGIN DATA.
 488 99
 489 99
 490 5.00
 491 END DATA.
 492
 493 MISSING VALUE X (99).
 494
 495 EXAMINE /x
 496         /PERCENTILES=HAVERAGE.
 497 ])
 498 AT_CHECK([pspp -o pspp.csv examine.sps])
 499 dnl Ignore output -- this is just a no-crash check.
 500 AT_CLEANUP
 501
 502 dnl Tests the trimmed mean calculation in the case
 503 dnl where the data is weighted towards the centre.
 504 AT_SETUP([EXAMINE -- trimmed mean])
 505 AT_DATA([examine.sps], [dnl
 506 DATA LIST LIST /X * C *.
 507 BEGIN DATA.
 508 1 1
 509 2 49
 510 3 2
 511 END DATA.
 512
 513 WEIGHT BY c.
 514
 515 EXAMINE
 516         x
 517         /STATISTICS=DESCRIPTIVES
 518         .
 519 ])
 520 AT_CHECK([pspp -o pspp.csv examine.sps])
 521 AT_CHECK([cat pspp.csv], [0], [dnl
 522 Table: Reading free-form data from INLINE.
 523 Variable,Format
 524 X,F8.0
 525 C,F8.0
 526
 527 Table: Case Processing Summary
 528 ,Cases,,,,,
 529 ,Valid,,Missing,,Total,
 530 ,N,Percent,N,Percent,N,Percent
 531 X,52.00,100%,.00,0%,52.00,100%
 532
 533 Table: Descriptives
 534 ,,,Statistic,Std. Error
 535 X,Mean,,2.02,.03
 536 ,95% Confidence Interval for Mean,Lower Bound,1.95,
 537 ,,Upper Bound,2.09,
 538 ,5% Trimmed Mean,,2.00,
 539 ,Median,,2.00,
 540 ,Variance,,.06,
 541 ,Std. Deviation,,.24,
 542 ,Minimum,,1.00,
 543 ,Maximum,,3.00,
 544 ,Range,,2.00,
 545 ,Interquartile Range,,.00,
 546 ,Skewness,,1.19,.33
 547 ,Kurtosis,,15.73,.65
 548 ])
 549 AT_CLEANUP
 550
 551 AT_SETUP([EXAMINE -- crash bug])
 552 AT_DATA([examine.sps], [dnl
 553 data list list /a * x * y *.
 554 begin data.
 555 3 1 3
 556 5 1 4
 557 7 2 3
 558 end data.
 559
 560 examine a by x by y
 561         /statistics=DESCRIPTIVES
 562         .
 563 ])
 564 AT_CHECK([pspp -o pspp.csv examine.sps])
 565 dnl Ignore output -- this is just a no-crash check.
 566 AT_CLEANUP
 567
 568 dnl Test that two consecutive EXAMINE commands don't crash PSPP.
 569 AT_SETUP([EXAMINE -- consecutive runs don't crash])
 570 AT_DATA([examine.sps], [dnl
 571 data list list /y * z *.
 572 begin data.
 573 6 4
 574 5 3
 575 7 6
 576 end data.
 577
 578 EXAMINE /VARIABLES= z BY y.
 579
 580 EXAMINE /VARIABLES= z.
 581 ])
 582 AT_CHECK([pspp -o pspp.csv examine.sps])
 583 dnl Ignore output -- this is just a no-crash check.
 584 AT_CLEANUP
 585
 586 dnl Test that /DESCRIPTIVES does not crash in presence of missing values.
 587 AT_SETUP([EXAMINE -- missing values don't crash])
 588 AT_DATA([examine.sps], [dnl
 589 data list list /x * y *.
 590 begin data.
 591 1 0
 592 2 0
 593 . 0
 594 3 1
 595 4 1
 596 end data.
 597 examine x by y /statistics=descriptives.
 598 ])
 599 AT_CHECK([pspp -o pspp.csv examine.sps])
 600 dnl Ignore output -- this is just a no-crash check.
 601 AT_CLEANUP
 602
 603 dnl Test that having only a single case doesn't crash.
 604 AT_SETUP([EXAMINE -- single case doesn't crash])
 605 AT_DATA([examine.sps], [dnl
 606 DATA LIST LIST /quality * .
 607 BEGIN DATA
 608 3
 609 END DATA
 610
 611
 612 EXAMINE
 613         quality
 614         /STATISTICS descriptives
 615         /PLOT = histogram
 616         .
 617 ])
 618 AT_CHECK([pspp -o pspp.csv examine.sps], [0], [ignore])
 619 dnl Ignore output -- this is just a no-crash check.
 620 AT_CLEANUP
 621
 622 dnl Test that all-missing data doesn't crash.
 623 AT_SETUP([EXAMINE -- all-missing data doesn't crash])
 624 AT_DATA([examine.sps], [dnl
 625 DATA LIST LIST /x *.
 626 BEGIN DATA.
 627 .
 628 .
 629 .
 630 .
 631 END DATA.
 632
 633 EXAMINE /x
 634         PLOT=HISTOGRAM BOXPLOT NPPLOT SPREADLEVEL(1) ALL
 635         /ID=x
 636         /STATISTICS = DESCRIPTIVES EXTREME (5) ALL
 637         /PERCENTILE=AEMPIRICAL
 638         .
 639 ])
 640 AT_CHECK([pspp -o pspp.csv examine.sps], [0], [ignore])
 641 dnl Ignore output -- this is just a no-crash check.
 642 AT_CLEANUP
 643
 644 dnl Test that big input doesn't crash (bug 11307).
 645 AT_SETUP([EXAMINE -- big input doesn't crash])
 646 AT_DATA([examine.sps], [dnl
 647 INPUT PROGRAM.
 648         LOOP #I=1 TO 50000.
 649                 COMPUTE X=NORMAL(10).
 650                 END CASE.
 651         END LOOP.
 652         END FILE.
 653 END INPUT PROGRAM.
 654
 655
 656 EXAMINE /x
 657         /STATISTICS=DESCRIPTIVES.
 658 ])
 659 AT_CHECK([pspp -o pspp.csv examine.sps])
 660 dnl Ignore output -- this is just a no-crash check.
 661 AT_CLEANUP
 662
 663 dnl Another test that big input doesn't crash.
 664 dnl The actual bug that this checks for has been lost.
 665 AT_SETUP([EXAMINE -- big input doesn't crash 2])
 666 AT_DATA([make-big-input.pl],
 667   [for ($i=0; $i<100000; $i++) { print "AB12\n" };
 668    for ($i=0; $i<100000; $i++) { print "AB04\n" };
 669 ])
 670 AT_CHECK([$PERL make-big-input.pl > large.txt])
 671 AT_DATA([examine.sps], [dnl
 672 DATA LIST FILE='large.txt' /S 1-2 (A) X 3 .
 673
 674
 675 AGGREGATE OUTFILE=* /BREAK=X /A=N.
 676
 677
 678 EXAMINE /A BY X.
 679 ])
 680 AT_CHECK([pspp -o pspp.csv examine.sps])
 681 dnl Ignore output -- this is just a no-crash check.
 682 AT_DATA([more-big-input.pl],
 683   [for ($i=0; $i<25000; $i++) { print "AB04\nAB12\n" };
 684 ])
 685 AT_CHECK([$PERL more-big-input.pl >> large.txt])
 686 AT_CHECK([pspp -o pspp.csv examine.sps])
 687 dnl Ignore output -- this is just a no-crash check.
 688 AT_CLEANUP
 689
 690
 691 dnl Test that the ID command works with non-numberic variables
 692 AT_SETUP([EXAMINE -- non-numeric ID])
 693
 694 AT_DATA([examine-id.sps], [dnl
 695 data list notable list /x * y (a12).
 696 begin data.
 697 1  one
 698 2  two
 699 3  three
 700 4  four
 701 5  five
 702 6  six
 703 7  seven
 704 8  eight
 705 9  nine
 706 10 ten
 707 11 eleven
 708 12 twelve
 709 30 thirty
 710 300 threehundred
 711 end data.
 712
 713 examine x
 714         /statistics = extreme
 715         /id = y
 716         /plot = boxplot
 717         .
 718 ])
 719
 720 AT_CHECK([pspp -O format=csv examine-id.sps], [0],
 721 [Table: Case Processing Summary
 722 ,Cases,,,,,
 723 ,Valid,,Missing,,Total,
 724 ,N,Percent,N,Percent,N,Percent
 725 x,14,100%,0,0%,14,100%
 726
 727 Table: Extreme Values
 728 ,,,y,Value
 729 x,Highest,1,threehundred,300.00
 730 ,,2,thirty      ,30.00
 731 ,,3,twelve      ,12.00
 732 ,,4,eleven      ,11.00
 733 ,,5,ten         ,10.00
 734 ,Lowest,1,one         ,1.00
 735 ,,2,two         ,2.00
 736 ,,3,three       ,3.00
 737 ,,4,four        ,4.00
 738 ,,5,five        ,5.00
 739 ])
 740
 741 AT_CLEANUP
 742
 743 dnl Test for a crash which happened on cleanup from a bad input syntax
 744 AT_SETUP([EXAMINE -- Bad Input])
 745
 746 AT_DATA([examine-bad.sps], [dnl
 747 data list list /h * g *.
 748 begin data.
 749 1 1
 750 2 1
 751 3 1
 752 4 1
 753 5 2
 754 6 2
 755 7 2
 756 8 2
 757 9 2
 758 end data.
 759
 760 EXAMINE
 761         /VARIABLES= h
 762         BY  g
 763         /STATISTICS = DESCRIPTIVES EXTREME
 764         /PLOT = lkajsdas
 765         .
 766 ])
 767
 768 AT_CHECK([pspp -o pspp.csv examine-bad.sps], [1], [ignore])
 769
 770 AT_CLEANUP
 771
 772
 773 dnl Check the MISSING=REPORT option
 774 AT_SETUP([EXAMINE -- MISSING=REPORT])
 775
 776
 777 AT_DATA([examine-report.sps], [dnl
 778 set format = F22.0.
 779 data list list /x * g *.
 780 begin data.
 781 1   1
 782 2   1
 783 3   1
 784 4   1
 785 5   1
 786 6   1
 787 7   1
 788 8   1
 789 9   1
 790 10   2
 791 20   2
 792 30   2
 793 40   2
 794 50   2
 795 60   2
 796 70   2
 797 80   2
 798 90   2
 799 101   9
 800 201   9
 801 301   9
 802 401   9
 803 501   99
 804 601   99
 805 701   99
 806 801   99
 807 901   99
 808 1001  .
 809 2002  .
 810 3003  .
 811 4004  .
 812 end data.
 813
 814 MISSING VALUES g (9, 99, 999).
 815
 816 EXAMINE
 817         /VARIABLES = x
 818         BY  g
 819         /STATISTICS = EXTREME
 820         /NOTOTAL
 821         /MISSING = REPORT.
 822 ])
 823
 824
 825 AT_CHECK([pspp -O format=csv examine-report.sps], [0], [dnl
 826 Table: Reading free-form data from INLINE.
 827 Variable,Format
 828 x,F8.0
 829 g,F8.0
 830
 831 Table: Case Processing Summary
 832 ,,Cases,,,,,
 833 ,,Valid,,Missing,,Total,
 834 ,g,N,Percent,N,Percent,N,Percent
 835 x,. (missing),4,100%,0,0%,4,100%
 836 ,1,9,100%,0,0%,9,100%
 837 ,2,9,100%,0,0%,9,100%
 838 ,9 (missing),4,100%,0,0%,4,100%
 839 ,99 (missing),5,100%,0,0%,5,100%
 840
 841 Table: Extreme Values
 842 ,g,,,Case Number,Value
 843 x,. (missing),Highest,1,31,4004
 844 ,,,2,30,3003
 845 ,,,3,29,2002
 846 ,,,4,28,1001
 847 ,,,5,0,0
 848 ,,Lowest,1,28,1001
 849 ,,,2,29,2002
 850 ,,,3,30,3003
 851 ,,,4,31,4004
 852 ,,,5,31,4004
 853 ,1,Highest,1,9,9
 854 ,,,2,8,8
 855 ,,,3,7,7
 856 ,,,4,6,6
 857 ,,,5,5,5
 858 ,,Lowest,1,1,1
 859 ,,,2,2,2
 860 ,,,3,3,3
 861 ,,,4,4,4
 862 ,,,5,5,5
 863 ,2,Highest,1,18,90
 864 ,,,2,17,80
 865 ,,,3,16,70
 866 ,,,4,15,60
 867 ,,,5,14,50
 868 ,,Lowest,1,10,10
 869 ,,,2,11,20
 870 ,,,3,12,30
 871 ,,,4,13,40
 872 ,,,5,14,50
 873 ,9 (missing),Highest,1,22,401
 874 ,,,2,21,301
 875 ,,,3,20,201
 876 ,,,4,19,101
 877 ,,,5,0,0
 878 ,,Lowest,1,19,101
 879 ,,,2,20,201
 880 ,,,3,21,301
 881 ,,,4,22,401
 882 ,,,5,22,401
 883 ,99 (missing),Highest,1,27,901
 884 ,,,2,26,801
 885 ,,,3,25,701
 886 ,,,4,24,601
 887 ,,,5,23,501
 888 ,,Lowest,1,23,501
 889 ,,,2,24,601
 890 ,,,3,25,701
 891 ,,,4,26,801
 892 ,,,5,27,901
 893 ])
 894
 895
 896 AT_CLEANUP
 897
 898
 899 dnl Run a test of the basic STATISTICS using a "real"
 900 dnl dataset and comparing with "real" results kindly
 901 dnl provided by Olaf Nöhring
 902 AT_SETUP([EXAMINE -- sample unweighted])
 903
 904 AT_DATA([sample.sps], [dnl
 905 set format = F22.4.
 906 DATA LIST notable LIST /X *
 907 BEGIN DATA.
 908 461.19000000
 909 466.38000000
 910 479.46000000
 911 480.10000000
 912 483.43000000
 913 488.30000000
 914 489.00000000
 915 491.62000000
 916 505.62000000
 917 511.30000000
 918 521.53000000
 919 526.70000000
 920 528.25000000
 921 538.70000000
 922 540.22000000
 923 540.58000000
 924 546.10000000
 925 548.17000000
 926 553.99000000
 927 566.21000000
 928 575.90000000
 929 584.38000000
 930 593.40000000
 931 357.05000000
 932 359.73000000
 933 360.48000000
 934 373.98000000
 935 374.13000000
 936 381.45000000
 937 383.72000000
 938 390.00000000
 939 400.34000000
 940 415.32000000
 941 415.91000000
 942 418.30000000
 943 421.03000000
 944 422.43000000
 945 426.93000000
 946 433.25000000
 947 436.89000000
 948 445.33000000
 949 446.33000000
 950 446.55000000
 951 456.44000000
 952 689.49000000
 953 691.92000000
 954 695.00000000
 955 695.36000000
 956 698.21000000
 957 699.46000000
 958 706.61000000
 959 710.69000000
 960 715.82000000
 961 715.82000000
 962 741.39000000
 963 752.27000000
 964 756.73000000
 965 757.74000000
 966 759.57000000
 967 796.07000000
 968 813.78000000
 969 817.25000000
 970 825.48000000
 971 831.28000000
 972 849.24000000
 973 890.00000000
 974 894.78000000
 975 935.65000000
 976 935.90000000
 977 945.90000000
 978 1012.8600000
 979 1022.6000000
 980 1061.8100000
 981 1063.5000000
 982 1077.2300000
 983 1151.6300000
 984 1355.2800000
 985 598.88000000
 986 606.91000000
 987 621.60000000
 988 624.80000000
 989 636.13000000
 990 637.38000000
 991 640.32000000
 992 649.35000000
 993 656.51000000
 994 662.55000000
 995 664.69000000
 996 106.22000000
 997 132.24000000
 998 174.76000000
 999 204.85000000
1000 264.93000000
1001 264.99000000
1002 269.84000000
1003 325.12000000
1004 331.67000000
1005 337.26000000
1006 347.68000000
1007 354.91000000
1008 END DATA.
1009
1010 EXAMINE
1011         x
1012         /STATISTICS=DESCRIPTIVES
1013         .
1014 ])
1015
1016 AT_CHECK([pspp -O format=csv sample.sps], [0], [dnl
1017 Table: Case Processing Summary
1018 ,Cases,,,,,
1019 ,Valid,,Missing,,Total,
1020 ,N,Percent,N,Percent,N,Percent
1021 X,100,100%,0,0%,100,100%
1022
1023 Table: Descriptives
1024 ,,,Statistic,Std. Error
1025 X,Mean,,587.6603,23.2665
1026 ,95% Confidence Interval for Mean,Lower Bound,541.4946,
1027 ,,Upper Bound,633.8260,
1028 ,5% Trimmed Mean,,579.7064,
1029 ,Median,,547.1350,
1030 ,Variance,,54132.8466,
1031 ,Std. Deviation,,232.6647,
1032 ,Minimum,,106.2200,
1033 ,Maximum,,1355.2800,
1034 ,Range,,1249.0600,
1035 ,Interquartile Range,,293.1575,
1036 ,Skewness,,.6331,.2414
1037 ,Kurtosis,,.5300,.4783
1038 ])
1039
1040 AT_CLEANUP
1041
1042
1043
1044 dnl Test for a crash which happened on bad input syntax
1045 AT_SETUP([EXAMINE -- Empty Parentheses])
1046
1047 AT_DATA([examine-empty-parens.sps], [dnl
1048 DATA LIST notable LIST /X *
1049 BEGIN DATA.
1050 2
1051 3
1052 END DATA.
1053
1054
1055 EXAMINE
1056         x
1057         /PLOT = SPREADLEVEL()
1058         .
1059 ])
1060
1061 AT_CHECK([pspp -o pspp.csv examine-empty-parens.sps], [1], [ignore])
1062
1063 AT_CLEANUP
1064
1065
1066
1067
1068 dnl Test for another crash which happened on bad input syntax
1069 AT_SETUP([EXAMINE -- Bad variable])
1070
1071 AT_DATA([examine-bad-variable.sps], [dnl
1072 data list list /h * g *.
1073 begin data.
1074 3 1
1075 4 1
1076 5 2
1077 end data.
1078
1079 EXAMINE
1080         /VARIABLES/ h
1081         BY  g
1082         .
1083 ])
1084
1085 AT_CHECK([pspp -o pspp.csv examine-bad-variable.sps], [1], [ignore])
1086
1087 AT_CLEANUP
1088
1089
1090
1091 dnl Test for yet another crash. This time for extremes vs. missing weight values.\0
1092 AT_SETUP([EXAMINE -- Extremes vs. Missing Weights])
1093
1094 AT_DATA([examine-missing-weights.sps], [dnl
1095 data list notable list /h * g *.
1096 begin data.
1097 3 1
1098 4 .
1099 5 1
1100 2 1
1101 end data.
1102
1103 WEIGHT BY g.
1104
1105 EXAMINE h
1106         /STATISTICS extreme(3)
1107         .
1108 ])
1109
1110 AT_CHECK([pspp -O format=csv  examine-missing-weights.sps], [0], [dnl
1111 "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."
1112
1113 Table: Case Processing Summary
1114 ,Cases,,,,,
1115 ,Valid,,Missing,,Total,
1116 ,N,Percent,N,Percent,N,Percent
1117 h,3.00,100%,.00,0%,3.00,100%
1118
1119 Table: Extreme Values
1120 ,,,Case Number,Value
1121 h,Highest,1,3,5.00
1122 ,,2,2,4.00
1123 ,,3,1,3.00
1124 ,Lowest,1,4,2.00
1125 ,,2,1,3.00
1126 ,,3,2,4.00
1127 ])
1128
1129 AT_CLEANUP
1130
1131
1132