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
  17 AT_BANNER([EXAMINE])
  18
  19 AT_SETUP([EXAMINE])
  20 AT_DATA([examine.sps], [
  21 DATA LIST LIST /QUALITY * W * BRAND * .
  22 BEGIN DATA
  23 3  1  1
  24 2  2  1
  25 1  2  1
  26 1  1  1
  27 4  1  1
  28 4  1  1
  29 5  1  2
  30 2  1  2
  31 4  4  2
  32 2  1  2
  33 3  1  2
  34 7  1  3
  35 4  2  3
  36 5  3  3
  37 3  1  3
  38 6  1  3
  39 END DATA
  40
  41 WEIGHT BY w.
  42
  43 VARIABLE LABELS brand   'Manufacturer'.
  44 VARIABLE LABELS quality 'Breaking Strain'.
  45
  46 VALUE LABELS /brand 1 'Aspeger' 2 'Bloggs' 3 'Charlies'.
  47
  48 LIST /FORMAT=NUMBERED.
  49
  50 EXAMINE
  51         quality BY brand
  52         /STATISTICS descriptives extreme(3)
  53         .
  54 ])
  55
  56
  57 dnl In the following data, only the extreme values have been checked.
  58 dnl The descriptives have been blindly pasted.
  59 AT_CHECK([pspp -O format=csv examine.sps], [0], [dnl
  60 Table: Reading free-form data from INLINE.
  61 Variable,Format
  62 QUALITY,F8.0
  63 W,F8.0
  64 BRAND,F8.0
  65
  66 Table: Data List
  67 Case Number,QUALITY,W,BRAND
  68 1,3.00,1.00,1.00
  69 2,2.00,2.00,1.00
  70 3,1.00,2.00,1.00
  71 4,1.00,1.00,1.00
  72 5,4.00,1.00,1.00
  73 6,4.00,1.00,1.00
  74 7,5.00,1.00,2.00
  75 8,2.00,1.00,2.00
  76 9,4.00,4.00,2.00
  77 10,2.00,1.00,2.00
  78 11,3.00,1.00,2.00
  79 12,7.00,1.00,3.00
  80 13,4.00,2.00,3.00
  81 14,5.00,3.00,3.00
  82 15,3.00,1.00,3.00
  83 16,6.00,1.00,3.00
  84
  85 Table: Case Processing Summary
  86 ,Cases,,,,,
  87 ,Valid,,Missing,,Total,
  88 ,N,Percent,N,Percent,N,Percent
  89 Breaking Strain,24.00,100%,.00,0%,24.00,100%
  90
  91 Table: Extreme Values
  92 ,,,Case Number,Value
  93 Breaking Strain,Highest,1,12,7.00
  94 ,,2,16,6.00
  95 ,,3,14,5.00
  96 ,Lowest,1,3,1.00
  97 ,,2,4,1.00
  98 ,,3,2,2.00
  99
 100 Table: Descriptives
 101 ,,,Statistic,Std. Error
 102 Breaking Strain,Mean,,3.54,.32
 103 ,95% Confidence Interval for Mean,Lower Bound,2.87,
 104 ,,Upper Bound,4.21,
 105 ,5% Trimmed Mean,,3.50,
 106 ,Median,,4.00,
 107 ,Variance,,2.52,
 108 ,Std. Deviation,,1.59,
 109 ,Minimum,,1.00,
 110 ,Maximum,,7.00,
 111 ,Range,,6.00,
 112 ,Interquartile Range,,2.75,
 113 ,Skewness,,.06,.47
 114 ,Kurtosis,,-.36,.92
 115
 116 Table: Case Processing Summary
 117 ,,Cases,,,,,
 118 ,,Valid,,Missing,,Total,
 119 ,Manufacturer,N,Percent,N,Percent,N,Percent
 120 Breaking Strain,Aspeger,8.00,100%,.00,0%,8.00,100%
 121 ,Bloggs,8.00,100%,.00,0%,8.00,100%
 122 ,Charlies,8.00,100%,.00,0%,8.00,100%
 123
 124 Table: Extreme Values
 125 ,Manufacturer,,,Case Number,Value
 126 Breaking Strain,Aspeger,Highest,1,6,4.00
 127 ,,,2,5,4.00
 128 ,,,3,1,3.00
 129 ,,Lowest,1,3,1.00
 130 ,,,2,4,1.00
 131 ,,,3,2,2.00
 132 ,Bloggs,Highest,1,7,5.00
 133 ,,,2,9,4.00
 134 ,,,3,11,3.00
 135 ,,Lowest,1,8,2.00
 136 ,,,2,10,2.00
 137 ,,,3,11,3.00
 138 ,Charlies,Highest,1,12,7.00
 139 ,,,2,16,6.00
 140 ,,,3,14,5.00
 141 ,,Lowest,1,15,3.00
 142 ,,,2,13,4.00
 143 ,,,3,14,5.00
 144
 145 Table: Descriptives
 146 ,Manufacturer,,,Statistic,Std. Error
 147 Breaking Strain,Aspeger,Mean,,2.25,.45
 148 ,,95% Confidence Interval for Mean,Lower Bound,1.18,
 149 ,,,Upper Bound,3.32,
 150 ,,5% Trimmed Mean,,2.22,
 151 ,,Median,,2.00,
 152 ,,Variance,,1.64,
 153 ,,Std. Deviation,,1.28,
 154 ,,Minimum,,1.00,
 155 ,,Maximum,,4.00,
 156 ,,Range,,3.00,
 157 ,,Interquartile Range,,2.75,
 158 ,,Skewness,,.47,.75
 159 ,,Kurtosis,,-1.55,1.48
 160 ,Bloggs,Mean,,3.50,.38
 161 ,,95% Confidence Interval for Mean,Lower Bound,2.61,
 162 ,,,Upper Bound,4.39,
 163 ,,5% Trimmed Mean,,3.50,
 164 ,,Median,,4.00,
 165 ,,Variance,,1.14,
 166 ,,Std. Deviation,,1.07,
 167 ,,Minimum,,2.00,
 168 ,,Maximum,,5.00,
 169 ,,Range,,3.00,
 170 ,,Interquartile Range,,1.75,
 171 ,,Skewness,,-.47,.75
 172 ,,Kurtosis,,-.83,1.48
 173 ,Charlies,Mean,,4.88,.44
 174 ,,95% Confidence Interval for Mean,Lower Bound,3.83,
 175 ,,,Upper Bound,5.92,
 176 ,,5% Trimmed Mean,,4.86,
 177 ,,Median,,5.00,
 178 ,,Variance,,1.55,
 179 ,,Std. Deviation,,1.25,
 180 ,,Minimum,,3.00,
 181 ,,Maximum,,7.00,
 182 ,,Range,,4.00,
 183 ,,Interquartile Range,,1.75,
 184 ,,Skewness,,.30,.75
 185 ,,Kurtosis,,.15,1.48
 186 ])
 187
 188 AT_CLEANUP
 189
 190 AT_SETUP([EXAMINE -- extremes])
 191 AT_DATA([examine.sps], [dnl
 192 data list free /V1 W
 193 begin data.
 194 1  1
 195 2  1
 196 3  2
 197 3  1
 198 4  1
 199 5  1
 200 6  1
 201 7  1
 202 8  1
 203 9  1
 204 10 1
 205 11 1
 206 12 1
 207 13 1
 208 14 1
 209 15 1
 210 16 1
 211 17 1
 212 18 2
 213 19 1
 214 20 1
 215 end data.
 216
 217 weight by w.
 218
 219 examine v1
 220  /statistics=extreme(6)
 221  .
 222 ])
 223
 224 AT_CHECK([pspp -O format=csv examine.sps], [0],[dnl
 225 Table: Case Processing Summary
 226 ,Cases,,,,,
 227 ,Valid,,Missing,,Total,
 228 ,N,Percent,N,Percent,N,Percent
 229 V1,23.00,100%,.00,0%,23.00,100%
 230
 231 Table: Extreme Values
 232 ,,,Case Number,Value
 233 V1,Highest,1,21,20.00
 234 ,,2,20,19.00
 235 ,,3,19,18.00
 236 ,,4,18,17.00
 237 ,,5,17,16.00
 238 ,,6,16,15.00
 239 ,Lowest,1,1,1.00
 240 ,,2,2,2.00
 241 ,,3,3,3.00
 242 ,,4,4,3.00
 243 ,,5,5,4.00
 244 ,,6,6,5.00
 245 ])
 246
 247 AT_CLEANUP
 248
 249
 250
 251 AT_SETUP([EXAMINE -- extremes with fractional weights])
 252 AT_DATA([extreme.sps], [dnl
 253 set format=F20.3.
 254 data list notable list /w * x *.
 255 begin data.
 256  0.88  300000
 257  0.86  320000
 258  0.98  480000
 259  0.93  960000
 260  1.35  960000
 261  1.31  960000
 262  0.88  960000
 263  0.88  1080000
 264  0.88  1080000
 265  0.95  1200000
 266  1.47  1200000
 267  0.93  1200000
 268  0.98  1320000
 269  1.31  1380000
 270  0.93  1440000
 271  0.88  1560000
 272  1.56  1560000
 273  1.47  1560000
 274 end data.
 275
 276 weight by w.
 277
 278
 279 EXAMINE
 280         x
 281         /STATISTICS = DESCRIPTIVES EXTREME (5)
 282         .
 283 ])
 284
 285 AT_CHECK([pspp -O format=csv  extreme.sps], [0], [dnl
 286 Table: Case Processing Summary
 287 ,Cases,,,,,
 288 ,Valid,,Missing,,Total,
 289 ,N,Percent,N,Percent,N,Percent
 290 x,19.430,100%,.000,0%,19.430,100%
 291
 292 Table: Extreme Values
 293 ,,,Case Number,Value
 294 x,Highest,1,18,1560000.000
 295 ,,2,17,1560000.000
 296 ,,3,16,1560000.000
 297 ,,4,15,1440000.000
 298 ,,5,14,1380000.000
 299 ,Lowest,1,1,300000.000
 300 ,,2,2,320000.000
 301 ,,3,3,480000.000
 302 ,,4,4,960000.000
 303 ,,5,5,960000.000
 304
 305 Table: Descriptives
 306 ,,,Statistic,Std. Error
 307 x,Mean,,1120010.293,86222.178
 308 ,95% Confidence Interval for Mean,Lower Bound,939166.693,
 309 ,,Upper Bound,1300853.894,
 310 ,5% Trimmed Mean,,1141017.899,
 311 ,Median,,1200000.000,
 312 ,Variance,,144447748124.869,
 313 ,Std. Deviation,,380062.821,
 314 ,Minimum,,300000.000,
 315 ,Maximum,,1560000.000,
 316 ,Range,,1260000.000,
 317 ,Interquartile Range,,467258.065,
 318 ,Skewness,,-.887,.519
 319 ,Kurtosis,,.340,1.005
 320 ])
 321
 322 AT_CLEANUP
 323
 324 dnl Test the PERCENTILES subcommand of the EXAMINE command.
 325 dnl In particular test that it behaves properly when there are only
 326 dnl a few cases.
 327 AT_SETUP([EXAMINE -- percentiles])
 328 AT_DATA([examine.sps], [dnl
 329 DATA LIST LIST /X *.
 330 BEGIN DATA.
 331 2.00
 332 8.00
 333 5.00
 334 END DATA.
 335
 336 EXAMINE /x
 337         /PERCENTILES=HAVERAGE.
 338
 339 EXAMINE /x
 340         /PERCENTILES=WAVERAGE.
 341
 342 EXAMINE /x
 343         /PERCENTILES=ROUND.
 344
 345 EXAMINE /x
 346         /PERCENTILES=EMPIRICAL.
 347
 348 EXAMINE /x
 349         /PERCENTILES=AEMPIRICAL.
 350 ])
 351 AT_CHECK([pspp -o pspp.csv examine.sps])
 352 AT_CHECK([cat pspp.csv], [0], [dnl
 353 Table: Reading free-form data from INLINE.
 354 Variable,Format
 355 X,F8.0
 356
 357 Table: Case Processing Summary
 358 ,Cases,,,,,
 359 ,Valid,,Missing,,Total,
 360 ,N,Percent,N,Percent,N,Percent
 361 X,3,100%,0,0%,3,100%
 362
 363 Table: Percentiles
 364 ,,Percentiles,,,,,,
 365 ,,5,10,25,50,75,90,95
 366 X,HAverage,.40,.80,2.00,5.00,8.00,8.00,8.00
 367 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 368
 369 Table: Case Processing Summary
 370 ,Cases,,,,,
 371 ,Valid,,Missing,,Total,
 372 ,N,Percent,N,Percent,N,Percent
 373 X,3,100%,0,0%,3,100%
 374
 375 Table: Percentiles
 376 ,,Percentiles,,,,,,
 377 ,,5,10,25,50,75,90,95
 378 X,Weighted Average,.30,.60,1.50,3.50,5.75,7.10,7.55
 379 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 380
 381 Table: Case Processing Summary
 382 ,Cases,,,,,
 383 ,Valid,,Missing,,Total,
 384 ,N,Percent,N,Percent,N,Percent
 385 X,3,100%,0,0%,3,100%
 386
 387 Table: Percentiles
 388 ,,Percentiles,,,,,,
 389 ,,5,10,25,50,75,90,95
 390 X,Rounded,.00,.00,2.00,5.00,5.00,8.00,8.00
 391 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 392
 393 Table: Case Processing Summary
 394 ,Cases,,,,,
 395 ,Valid,,Missing,,Total,
 396 ,N,Percent,N,Percent,N,Percent
 397 X,3,100%,0,0%,3,100%
 398
 399 Table: Percentiles
 400 ,,Percentiles,,,,,,
 401 ,,5,10,25,50,75,90,95
 402 X,Empirical,2.00,2.00,2.00,5.00,8.00,8.00,8.00
 403 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 404
 405 Table: Case Processing Summary
 406 ,Cases,,,,,
 407 ,Valid,,Missing,,Total,
 408 ,N,Percent,N,Percent,N,Percent
 409 X,3,100%,0,0%,3,100%
 410
 411 Table: Percentiles
 412 ,,Percentiles,,,,,,
 413 ,,5,10,25,50,75,90,95
 414 X,Empirical with averaging,2.00,2.00,2.00,5.00,8.00,8.00,8.00
 415 ,Tukey's Hinges,,,3.50,5.00,6.50,,
 416 ])
 417 AT_CLEANUP
 418
 419 AT_SETUP([EXAMINE -- missing values])
 420 AT_DATA([examine.sps], [dnl
 421 DATA LIST LIST /x * y *.
 422 BEGIN DATA.
 423 1   1
 424 2   1
 425 3   1
 426 4   1
 427 5   2
 428 6   2
 429 .   2
 430 END DATA
 431
 432 EXAMINE /x by y
 433         /MISSING = PAIRWISE
 434         .
 435 ])
 436 AT_CHECK([pspp -o pspp.csv examine.sps])
 437 AT_CHECK([cat pspp.csv], [0], [dnl
 438 Table: Reading free-form data from INLINE.
 439 Variable,Format
 440 x,F8.0
 441 y,F8.0
 442
 443 Table: Case Processing Summary
 444 ,Cases,,,,,
 445 ,Valid,,Missing,,Total,
 446 ,N,Percent,N,Percent,N,Percent
 447 x,6,85.7143%,1,14.2857%,7,100%
 448
 449 Table: Case Processing Summary
 450 ,,Cases,,,,,
 451 ,,Valid,,Missing,,Total,
 452 ,y,N,Percent,N,Percent,N,Percent
 453 x,1.00,4,100%,0,0%,4,100%
 454 ,2.00,2,66.6667%,1,33.3333%,3,100%
 455 ])
 456 AT_CLEANUP
 457
 458
 459 AT_SETUP([EXAMINE -- user missing values])
 460 AT_DATA([examine-m.sps], [dnl
 461 DATA LIST notable LIST /x * y *.
 462 BEGIN DATA.
 463 1                   2
 464 9999999999          2
 465 9999999999          99
 466 END DATA.
 467
 468 MISSING VALUES x (9999999999).
 469 MISSING VALUES y (99).
 470
 471 EXAMINE
 472         /VARIABLES= x y
 473         /MISSING=PAIRWISE.
 474 ])
 475 AT_CHECK([pspp -O format=csv examine-m.sps], [0], [dnl
 476 Table: Case Processing Summary
 477 ,Cases,,,,,
 478 ,Valid,,Missing,,Total,
 479 ,N,Percent,N,Percent,N,Percent
 480 x,1,33.3333%,2,66.6667%,3,100%
 481 y,2,66.6667%,1,33.3333%,3,100%
 482 ])
 483 AT_CLEANUP
 484
 485 AT_SETUP([EXAMINE -- missing values and percentiles])
 486 AT_DATA([examine.sps], [dnl
 487 DATA LIST LIST /X *.
 488 BEGIN DATA.
 489 99
 490 99
 491 5.00
 492 END DATA.
 493
 494 MISSING VALUE X (99).
 495
 496 EXAMINE /x
 497         /PERCENTILES=HAVERAGE.
 498 ])
 499 AT_CHECK([pspp -o pspp.csv examine.sps])
 500 dnl Ignore output -- this is just a no-crash check.
 501 AT_CLEANUP
 502
 503 dnl Tests the trimmed mean calculation in the case
 504 dnl where the data is weighted towards the centre.
 505 AT_SETUP([EXAMINE -- trimmed mean])
 506 AT_DATA([examine.sps], [dnl
 507 DATA LIST LIST /X * C *.
 508 BEGIN DATA.
 509 1 1
 510 2 49
 511 3 2
 512 END DATA.
 513
 514 WEIGHT BY c.
 515
 516 EXAMINE
 517         x
 518         /STATISTICS=DESCRIPTIVES
 519         .
 520 ])
 521 AT_CHECK([pspp -o pspp.csv examine.sps])
 522 AT_CHECK([cat pspp.csv], [0], [dnl
 523 Table: Reading free-form data from INLINE.
 524 Variable,Format
 525 X,F8.0
 526 C,F8.0
 527
 528 Table: Case Processing Summary
 529 ,Cases,,,,,
 530 ,Valid,,Missing,,Total,
 531 ,N,Percent,N,Percent,N,Percent
 532 X,52.00,100%,.00,0%,52.00,100%
 533
 534 Table: Descriptives
 535 ,,,Statistic,Std. Error
 536 X,Mean,,2.02,.03
 537 ,95% Confidence Interval for Mean,Lower Bound,1.95,
 538 ,,Upper Bound,2.09,
 539 ,5% Trimmed Mean,,2.00,
 540 ,Median,,2.00,
 541 ,Variance,,.06,
 542 ,Std. Deviation,,.24,
 543 ,Minimum,,1.00,
 544 ,Maximum,,3.00,
 545 ,Range,,2.00,
 546 ,Interquartile Range,,.00,
 547 ,Skewness,,1.19,.33
 548 ,Kurtosis,,15.73,.65
 549 ])
 550 AT_CLEANUP
 551
 552 AT_SETUP([EXAMINE -- crash bug])
 553 AT_DATA([examine.sps], [dnl
 554 data list list /a * x * y *.
 555 begin data.
 556 3 1 3
 557 5 1 4
 558 7 2 3
 559 end data.
 560
 561 examine a by x by y
 562         /statistics=DESCRIPTIVES
 563         .
 564 ])
 565 AT_CHECK([pspp -o pspp.csv examine.sps])
 566 dnl Ignore output -- this is just a no-crash check.
 567 AT_CLEANUP
 568
 569 dnl Test that two consecutive EXAMINE commands don't crash PSPP.
 570 AT_SETUP([EXAMINE -- consecutive runs don't crash])
 571 AT_DATA([examine.sps], [dnl
 572 data list list /y * z *.
 573 begin data.
 574 6 4
 575 5 3
 576 7 6
 577 end data.
 578
 579 EXAMINE /VARIABLES= z BY y.
 580
 581 EXAMINE /VARIABLES= z.
 582 ])
 583 AT_CHECK([pspp -o pspp.csv examine.sps])
 584 dnl Ignore output -- this is just a no-crash check.
 585 AT_CLEANUP
 586
 587 dnl Test that /DESCRIPTIVES does not crash in presence of missing values.
 588 AT_SETUP([EXAMINE -- missing values don't crash])
 589 AT_DATA([examine.sps], [dnl
 590 data list list /x * y *.
 591 begin data.
 592 1 0
 593 2 0
 594 . 0
 595 3 1
 596 4 1
 597 end data.
 598 examine x by y /statistics=descriptives.
 599 ])
 600 AT_CHECK([pspp -o pspp.csv examine.sps])
 601 dnl Ignore output -- this is just a no-crash check.
 602 AT_CLEANUP
 603
 604 dnl Test that having only a single case doesn't crash.
 605 AT_SETUP([EXAMINE -- single case doesn't crash])
 606 AT_DATA([examine.sps], [dnl
 607 DATA LIST LIST /quality * .
 608 BEGIN DATA
 609 3
 610 END DATA
 611
 612
 613 EXAMINE
 614         quality
 615         /STATISTICS descriptives
 616         /PLOT = histogram
 617         .
 618 ])
 619 AT_CHECK([pspp -o pspp.csv examine.sps], [0], [ignore])
 620 dnl Ignore output -- this is just a no-crash check.
 621 AT_CLEANUP
 622
 623 dnl Test that all-missing data doesn't crash.
 624 AT_SETUP([EXAMINE -- all-missing data doesn't crash])
 625 AT_DATA([examine.sps], [dnl
 626 DATA LIST LIST /x *.
 627 BEGIN DATA.
 628 .
 629 .
 630 .
 631 .
 632 END DATA.
 633
 634 EXAMINE /x
 635         PLOT=HISTOGRAM BOXPLOT NPPLOT SPREADLEVEL(1) ALL
 636         /ID=x
 637         /STATISTICS = DESCRIPTIVES EXTREME (5) ALL
 638         /PERCENTILE=AEMPIRICAL
 639         .
 640 ])
 641 AT_CHECK([pspp -o pspp.csv examine.sps], [0], [ignore])
 642 dnl Ignore output -- this is just a no-crash check.
 643 AT_CLEANUP
 644
 645 dnl Test that big input doesn't crash (bug 11307).
 646 AT_SETUP([EXAMINE -- big input doesn't crash])
 647 AT_DATA([examine.sps], [dnl
 648 INPUT PROGRAM.
 649         LOOP #I=1 TO 50000.
 650                 COMPUTE X=NORMAL(10).
 651                 END CASE.
 652         END LOOP.
 653         END FILE.
 654 END INPUT PROGRAM.
 655
 656
 657 EXAMINE /x
 658         /STATISTICS=DESCRIPTIVES.
 659 ])
 660 AT_CHECK([pspp -o pspp.csv examine.sps])
 661 dnl Ignore output -- this is just a no-crash check.
 662 AT_CLEANUP
 663
 664 dnl Another test that big input doesn't crash.
 665 dnl The actual bug that this checks for has been lost.
 666 AT_SETUP([EXAMINE -- big input doesn't crash 2])
 667 AT_DATA([make-big-input.pl],
 668   [for ($i=0; $i<100000; $i++) { print "AB12\n" };
 669    for ($i=0; $i<100000; $i++) { print "AB04\n" };
 670 ])
 671 AT_CHECK([$PERL make-big-input.pl > large.txt])
 672 AT_DATA([examine.sps], [dnl
 673 DATA LIST FILE='large.txt' /S 1-2 (A) X 3 .
 674
 675
 676 AGGREGATE OUTFILE=* /BREAK=X /A=N.
 677
 678
 679 EXAMINE /A BY X.
 680 ])
 681 AT_CHECK([pspp -o pspp.csv examine.sps])
 682 dnl Ignore output -- this is just a no-crash check.
 683 AT_DATA([more-big-input.pl],
 684   [for ($i=0; $i<25000; $i++) { print "AB04\nAB12\n" };
 685 ])
 686 AT_CHECK([$PERL more-big-input.pl >> large.txt])
 687 AT_CHECK([pspp -o pspp.csv examine.sps])
 688 dnl Ignore output -- this is just a no-crash check.
 689 AT_CLEANUP
 690
 691
 692 dnl Test that the ID command works with non-numberic variables
 693 AT_SETUP([EXAMINE -- non-numeric ID])
 694
 695 AT_DATA([examine-id.sps], [dnl
 696 data list notable list /x * y (a12).
 697 begin data.
 698 1  one
 699 2  two
 700 3  three
 701 4  four
 702 5  five
 703 6  six
 704 7  seven
 705 8  eight
 706 9  nine
 707 10 ten
 708 11 eleven
 709 12 twelve
 710 30 thirty
 711 300 threehundred
 712 end data.
 713
 714 examine x
 715         /statistics = extreme
 716         /id = y
 717         /plot = boxplot
 718         .
 719 ])
 720
 721 AT_CHECK([pspp -O format=csv examine-id.sps], [0],
 722 [Table: Case Processing Summary
 723 ,Cases,,,,,
 724 ,Valid,,Missing,,Total,
 725 ,N,Percent,N,Percent,N,Percent
 726 x,14,100%,0,0%,14,100%
 727
 728 Table: Extreme Values
 729 ,,,y,Value
 730 x,Highest,1,threehundred,300.00
 731 ,,2,thirty      ,30.00
 732 ,,3,twelve      ,12.00
 733 ,,4,eleven      ,11.00
 734 ,,5,ten         ,10.00
 735 ,Lowest,1,one         ,1.00
 736 ,,2,two         ,2.00
 737 ,,3,three       ,3.00
 738 ,,4,four        ,4.00
 739 ,,5,five        ,5.00
 740 ])
 741
 742 AT_CLEANUP
 743
 744 dnl Test for a crash which happened on cleanup from a bad input syntax
 745 AT_SETUP([EXAMINE -- Bad Input])
 746
 747 AT_DATA([examine-bad.sps], [dnl
 748 data list list /h * g *.
 749 begin data.
 750 1 1
 751 2 1
 752 3 1
 753 4 1
 754 5 2
 755 6 2
 756 7 2
 757 8 2
 758 9 2
 759 end data.
 760
 761 EXAMINE
 762         /VARIABLES= h
 763         BY  g
 764         /STATISTICS = DESCRIPTIVES EXTREME
 765         /PLOT = lkajsdas
 766         .
 767 ])
 768
 769 AT_CHECK([pspp -o pspp.csv examine-bad.sps], [1], [ignore])
 770
 771 AT_CLEANUP
 772
 773
 774 dnl Check the MISSING=REPORT option
 775 AT_SETUP([EXAMINE -- MISSING=REPORT])
 776
 777
 778 AT_DATA([examine-report.sps], [dnl
 779 set format = F22.0.
 780 data list list /x * g *.
 781 begin data.
 782 1   1
 783 2   1
 784 3   1
 785 4   1
 786 5   1
 787 6   1
 788 7   1
 789 8   1
 790 9   1
 791 10   2
 792 20   2
 793 30   2
 794 40   2
 795 50   2
 796 60   2
 797 70   2
 798 80   2
 799 90   2
 800 101   9
 801 201   9
 802 301   9
 803 401   9
 804 501   99
 805 601   99
 806 701   99
 807 801   99
 808 901   99
 809 1001  .
 810 2002  .
 811 3003  .
 812 4004  .
 813 end data.
 814
 815 MISSING VALUES g (9, 99, 999).
 816
 817 EXAMINE
 818         /VARIABLES = x
 819         BY  g
 820         /STATISTICS = EXTREME
 821         /NOTOTAL
 822         /MISSING = REPORT.
 823 ])
 824
 825
 826 AT_CHECK([pspp -O format=csv examine-report.sps], [0], [dnl
 827 Table: Reading free-form data from INLINE.
 828 Variable,Format
 829 x,F8.0
 830 g,F8.0
 831
 832 Table: Case Processing Summary
 833 ,,Cases,,,,,
 834 ,,Valid,,Missing,,Total,
 835 ,g,N,Percent,N,Percent,N,Percent
 836 x,. (missing),4,100%,0,0%,4,100%
 837 ,1,9,100%,0,0%,9,100%
 838 ,2,9,100%,0,0%,9,100%
 839 ,9 (missing),4,100%,0,0%,4,100%
 840 ,99 (missing),5,100%,0,0%,5,100%
 841
 842 Table: Extreme Values
 843 ,g,,,Case Number,Value
 844 x,. (missing),Highest,1,31,4004
 845 ,,,2,30,3003
 846 ,,,3,29,2002
 847 ,,,4,28,1001
 848 ,,,5,0,0
 849 ,,Lowest,1,28,1001
 850 ,,,2,29,2002
 851 ,,,3,30,3003
 852 ,,,4,31,4004
 853 ,,,5,31,4004
 854 ,1,Highest,1,9,9
 855 ,,,2,8,8
 856 ,,,3,7,7
 857 ,,,4,6,6
 858 ,,,5,5,5
 859 ,,Lowest,1,1,1
 860 ,,,2,2,2
 861 ,,,3,3,3
 862 ,,,4,4,4
 863 ,,,5,5,5
 864 ,2,Highest,1,18,90
 865 ,,,2,17,80
 866 ,,,3,16,70
 867 ,,,4,15,60
 868 ,,,5,14,50
 869 ,,Lowest,1,10,10
 870 ,,,2,11,20
 871 ,,,3,12,30
 872 ,,,4,13,40
 873 ,,,5,14,50
 874 ,9 (missing),Highest,1,22,401
 875 ,,,2,21,301
 876 ,,,3,20,201
 877 ,,,4,19,101
 878 ,,,5,0,0
 879 ,,Lowest,1,19,101
 880 ,,,2,20,201
 881 ,,,3,21,301
 882 ,,,4,22,401
 883 ,,,5,22,401
 884 ,99 (missing),Highest,1,27,901
 885 ,,,2,26,801
 886 ,,,3,25,701
 887 ,,,4,24,601
 888 ,,,5,23,501
 889 ,,Lowest,1,23,501
 890 ,,,2,24,601
 891 ,,,3,25,701
 892 ,,,4,26,801
 893 ,,,5,27,901
 894 ])
 895
 896
 897 AT_CLEANUP
 898
 899
 900 dnl Run a test of the basic STATISTICS using a "real"
 901 dnl dataset and comparing with "real" results kindly
 902 dnl provided by Olaf Nöhring
 903 AT_SETUP([EXAMINE -- sample unweighted])
 904
 905 AT_DATA([sample.sps], [dnl
 906 set format = F22.4.
 907 DATA LIST notable LIST /X *
 908 BEGIN DATA.
 909 461.19000000
 910 466.38000000
 911 479.46000000
 912 480.10000000
 913 483.43000000
 914 488.30000000
 915 489.00000000
 916 491.62000000
 917 505.62000000
 918 511.30000000
 919 521.53000000
 920 526.70000000
 921 528.25000000
 922 538.70000000
 923 540.22000000
 924 540.58000000
 925 546.10000000
 926 548.17000000
 927 553.99000000
 928 566.21000000
 929 575.90000000
 930 584.38000000
 931 593.40000000
 932 357.05000000
 933 359.73000000
 934 360.48000000
 935 373.98000000
 936 374.13000000
 937 381.45000000
 938 383.72000000
 939 390.00000000
 940 400.34000000
 941 415.32000000
 942 415.91000000
 943 418.30000000
 944 421.03000000
 945 422.43000000
 946 426.93000000
 947 433.25000000
 948 436.89000000
 949 445.33000000
 950 446.33000000
 951 446.55000000
 952 456.44000000
 953 689.49000000
 954 691.92000000
 955 695.00000000
 956 695.36000000
 957 698.21000000
 958 699.46000000
 959 706.61000000
 960 710.69000000
 961 715.82000000
 962 715.82000000
 963 741.39000000
 964 752.27000000
 965 756.73000000
 966 757.74000000
 967 759.57000000
 968 796.07000000
 969 813.78000000
 970 817.25000000
 971 825.48000000
 972 831.28000000
 973 849.24000000
 974 890.00000000
 975 894.78000000
 976 935.65000000
 977 935.90000000
 978 945.90000000
 979 1012.8600000
 980 1022.6000000
 981 1061.8100000
 982 1063.5000000
 983 1077.2300000
 984 1151.6300000
 985 1355.2800000
 986 598.88000000
 987 606.91000000
 988 621.60000000
 989 624.80000000
 990 636.13000000
 991 637.38000000
 992 640.32000000
 993 649.35000000
 994 656.51000000
 995 662.55000000
 996 664.69000000
 997 106.22000000
 998 132.24000000
 999 174.76000000
1000 204.85000000
1001 264.93000000
1002 264.99000000
1003 269.84000000
1004 325.12000000
1005 331.67000000
1006 337.26000000
1007 347.68000000
1008 354.91000000
1009 END DATA.
1010
1011 EXAMINE
1012         x
1013         /STATISTICS=DESCRIPTIVES
1014         .
1015 ])
1016
1017 AT_CHECK([pspp -O format=csv sample.sps], [0], [dnl
1018 Table: Case Processing Summary
1019 ,Cases,,,,,
1020 ,Valid,,Missing,,Total,
1021 ,N,Percent,N,Percent,N,Percent
1022 X,100,100%,0,0%,100,100%
1023
1024 Table: Descriptives
1025 ,,,Statistic,Std. Error
1026 X,Mean,,587.6603,23.2665
1027 ,95% Confidence Interval for Mean,Lower Bound,541.4946,
1028 ,,Upper Bound,633.8260,
1029 ,5% Trimmed Mean,,579.7064,
1030 ,Median,,547.1350,
1031 ,Variance,,54132.8466,
1032 ,Std. Deviation,,232.6647,
1033 ,Minimum,,106.2200,
1034 ,Maximum,,1355.2800,
1035 ,Range,,1249.0600,
1036 ,Interquartile Range,,293.1575,
1037 ,Skewness,,.6331,.2414
1038 ,Kurtosis,,.5300,.4783
1039 ])
1040
1041 AT_CLEANUP
1042
1043
1044
1045 dnl Test for a crash which happened on bad input syntax
1046 AT_SETUP([EXAMINE -- Empty Parentheses])
1047
1048 AT_DATA([examine-empty-parens.sps], [dnl
1049 DATA LIST notable LIST /X *
1050 BEGIN DATA.
1051 2
1052 3
1053 END DATA.
1054
1055
1056 EXAMINE
1057         x
1058         /PLOT = SPREADLEVEL()
1059         .
1060 ])
1061
1062 AT_CHECK([pspp -o pspp.csv examine-empty-parens.sps], [1], [ignore])
1063
1064 AT_CLEANUP
1065
1066
1067
1068
1069 dnl Test for another crash which happened on bad input syntax
1070 AT_SETUP([EXAMINE -- Bad variable])
1071
1072 AT_DATA([examine-bad-variable.sps], [dnl
1073 data list list /h * g *.
1074 begin data.
1075 3 1
1076 4 1
1077 5 2
1078 end data.
1079
1080 EXAMINE
1081         /VARIABLES/ h
1082         BY  g
1083         .
1084 ])
1085
1086 AT_CHECK([pspp -o pspp.csv examine-bad-variable.sps], [1], [ignore])
1087
1088 AT_CLEANUP
1089
1090
1091
1092 dnl Test for yet another crash. This time for extremes vs. missing weight values.\0
1093 AT_SETUP([EXAMINE -- Extremes vs. Missing Weights])
1094
1095 AT_DATA([examine-missing-weights.sps], [dnl
1096 data list notable list /h * g *.
1097 begin data.
1098 3 1
1099 4 .
1100 5 1
1101 2 1
1102 end data.
1103
1104 WEIGHT BY g.
1105
1106 EXAMINE h
1107         /STATISTICS extreme(3)
1108         .
1109 ])
1110
1111 AT_CHECK([pspp -O format=csv  examine-missing-weights.sps], [0], [dnl
1112 "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."
1113
1114 Table: Case Processing Summary
1115 ,Cases,,,,,
1116 ,Valid,,Missing,,Total,
1117 ,N,Percent,N,Percent,N,Percent
1118 h,3.00,100%,.00,0%,3.00,100%
1119
1120 Table: Extreme Values
1121 ,,,Case Number,Value
1122 h,Highest,1,3,5.00
1123 ,,2,2,4.00
1124 ,,3,1,3.00
1125 ,Lowest,1,4,2.00
1126 ,,2,1,3.00
1127 ,,3,2,4.00
1128 ])
1129
1130 AT_CLEANUP
1131
1132
1133