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

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