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 PLOT=HISTOGRAM.
 619 ])
 620 AT_CHECK([pspp -o pspp.csv examine.sps], [0], [ignore])
 621 dnl Ignore output -- this is just a no-crash check.
 622 AT_CLEANUP
 623
 624 dnl Test that big input doesn't crash (bug 11307).
 625 AT_SETUP([EXAMINE -- big input doesn't crash])
 626 AT_DATA([examine.sps], [dnl
 627 INPUT PROGRAM.
 628         LOOP #I=1 TO 50000.
 629                 COMPUTE X=NORMAL(10).
 630                 END CASE.
 631         END LOOP.
 632         END FILE.
 633 END INPUT PROGRAM.
 634
 635
 636 EXAMINE /x
 637         /STATISTICS=DESCRIPTIVES.
 638 ])
 639 AT_CHECK([pspp -o pspp.csv examine.sps])
 640 dnl Ignore output -- this is just a no-crash check.
 641 AT_CLEANUP
 642
 643 dnl Another test that big input doesn't crash.
 644 dnl The actual bug that this checks for has been lost.
 645 AT_SETUP([EXAMINE -- big input doesn't crash 2])
 646 AT_DATA([make-big-input.pl],
 647   [for ($i=0; $i<100000; $i++) { print "AB12\n" };
 648    for ($i=0; $i<100000; $i++) { print "AB04\n" };
 649 ])
 650 AT_CHECK([$PERL make-big-input.pl > large.txt])
 651 AT_DATA([examine.sps], [dnl
 652 DATA LIST FILE='large.txt' /S 1-2 (A) X 3 .
 653
 654
 655 AGGREGATE OUTFILE=* /BREAK=X /A=N.
 656
 657
 658 EXAMINE /A BY X.
 659 ])
 660 AT_CHECK([pspp -o pspp.csv examine.sps])
 661 dnl Ignore output -- this is just a no-crash check.
 662 AT_DATA([more-big-input.pl],
 663   [for ($i=0; $i<25000; $i++) { print "AB04\nAB12\n" };
 664 ])
 665 AT_CHECK([$PERL more-big-input.pl >> large.txt])
 666 AT_CHECK([pspp -o pspp.csv examine.sps])
 667 dnl Ignore output -- this is just a no-crash check.
 668 AT_CLEANUP
 669
 670
 671 dnl Test that the ID command works with non-numberic variables
 672 AT_SETUP([EXAMINE -- non-numeric ID])
 673
 674 AT_DATA([examine-id.sps], [dnl
 675 data list notable list /x * y (a12).
 676 begin data.
 677 1  one
 678 2  two
 679 3  three
 680 4  four
 681 5  five
 682 6  six
 683 7  seven
 684 8  eight
 685 9  nine
 686 10 ten
 687 11 eleven
 688 12 twelve
 689 30 thirty
 690 300 threehundred
 691 end data.
 692
 693 examine x
 694         /statistics = extreme
 695         /id = y
 696         /plot = boxplot
 697         .
 698 ])
 699
 700 AT_CHECK([pspp -O format=csv examine-id.sps], [0],
 701 [Table: Case Processing Summary
 702 ,Cases,,,,,
 703 ,Valid,,Missing,,Total,
 704 ,N,Percent,N,Percent,N,Percent
 705 x,14,100%,0,0%,14,100%
 706
 707 Table: Extreme Values
 708 ,,,y,Value
 709 x,Highest,1,threehundred,300.00
 710 ,,2,thirty      ,30.00
 711 ,,3,twelve      ,12.00
 712 ,,4,eleven      ,11.00
 713 ,,5,ten         ,10.00
 714 ,Lowest,1,one         ,1.00
 715 ,,2,two         ,2.00
 716 ,,3,three       ,3.00
 717 ,,4,four        ,4.00
 718 ,,5,five        ,5.00
 719 ])
 720
 721 AT_CLEANUP
 722
 723 dnl Test for a crash which happened on cleanup from a bad input syntax
 724 AT_SETUP([EXAMINE -- Bad Input])
 725
 726 AT_DATA([examine-bad.sps], [dnl
 727 data list list /h * g *.
 728 begin data.
 729 1 1
 730 2 1
 731 3 1
 732 4 1
 733 5 2
 734 6 2
 735 7 2
 736 8 2
 737 9 2
 738 end data.
 739
 740 EXAMINE
 741         /VARIABLES= h
 742         BY  g
 743         /STATISTICS = DESCRIPTIVES EXTREME
 744         /PLOT = lkajsdas
 745         .
 746 ])
 747
 748 AT_CHECK([pspp -o pspp.csv examine-bad.sps], [1], [ignore])
 749
 750 AT_CLEANUP
 751
 752
 753 dnl Check the MISSING=REPORT option
 754 AT_SETUP([EXAMINE -- MISSING=REPORT])
 755
 756
 757 AT_DATA([examine-report.sps], [dnl
 758 set format = F22.0.
 759 data list list /x * g *.
 760 begin data.
 761 1   1
 762 2   1
 763 3   1
 764 4   1
 765 5   1
 766 6   1
 767 7   1
 768 8   1
 769 9   1
 770 10   2
 771 20   2
 772 30   2
 773 40   2
 774 50   2
 775 60   2
 776 70   2
 777 80   2
 778 90   2
 779 101   9
 780 201   9
 781 301   9
 782 401   9
 783 501   99
 784 601   99
 785 701   99
 786 801   99
 787 901   99
 788 1001  .
 789 2002  .
 790 3003  .
 791 4004  .
 792 end data.
 793
 794 MISSING VALUES g (9, 99, 999).
 795
 796 EXAMINE
 797         /VARIABLES = x
 798         BY  g
 799         /STATISTICS = EXTREME
 800         /NOTOTAL
 801         /MISSING = REPORT.
 802 ])
 803
 804
 805 AT_CHECK([pspp -O format=csv examine-report.sps], [0], [dnl
 806 Table: Reading free-form data from INLINE.
 807 Variable,Format
 808 x,F8.0
 809 g,F8.0
 810
 811 Table: Case Processing Summary
 812 ,,Cases,,,,,
 813 ,,Valid,,Missing,,Total,
 814 ,g,N,Percent,N,Percent,N,Percent
 815 x,. (missing),4,100%,0,0%,4,100%
 816 ,1,9,100%,0,0%,9,100%
 817 ,2,9,100%,0,0%,9,100%
 818 ,9 (missing),4,100%,0,0%,4,100%
 819 ,99 (missing),5,100%,0,0%,5,100%
 820
 821 Table: Extreme Values
 822 ,g,,,Case Number,Value
 823 x,. (missing),Highest,1,31,4004
 824 ,,,2,30,3003
 825 ,,,3,29,2002
 826 ,,,4,28,1001
 827 ,,,5,0,0
 828 ,,Lowest,1,28,1001
 829 ,,,2,29,2002
 830 ,,,3,30,3003
 831 ,,,4,31,4004
 832 ,,,5,31,4004
 833 ,1,Highest,1,9,9
 834 ,,,2,8,8
 835 ,,,3,7,7
 836 ,,,4,6,6
 837 ,,,5,5,5
 838 ,,Lowest,1,1,1
 839 ,,,2,2,2
 840 ,,,3,3,3
 841 ,,,4,4,4
 842 ,,,5,5,5
 843 ,2,Highest,1,18,90
 844 ,,,2,17,80
 845 ,,,3,16,70
 846 ,,,4,15,60
 847 ,,,5,14,50
 848 ,,Lowest,1,10,10
 849 ,,,2,11,20
 850 ,,,3,12,30
 851 ,,,4,13,40
 852 ,,,5,14,50
 853 ,9 (missing),Highest,1,22,401
 854 ,,,2,21,301
 855 ,,,3,20,201
 856 ,,,4,19,101
 857 ,,,5,0,0
 858 ,,Lowest,1,19,101
 859 ,,,2,20,201
 860 ,,,3,21,301
 861 ,,,4,22,401
 862 ,,,5,22,401
 863 ,99 (missing),Highest,1,27,901
 864 ,,,2,26,801
 865 ,,,3,25,701
 866 ,,,4,24,601
 867 ,,,5,23,501
 868 ,,Lowest,1,23,501
 869 ,,,2,24,601
 870 ,,,3,25,701
 871 ,,,4,26,801
 872 ,,,5,27,901
 873 ])
 874
 875
 876 AT_CLEANUP
 877
 878
 879 dnl Run a test of the basic STATISTICS using a "real"
 880 dnl dataset and comparing with "real" results kindly
 881 dnl provided by Olaf Nöhring
 882 AT_SETUP([EXAMINE -- sample unweighted])
 883
 884 AT_DATA([sample.sps], [dnl
 885 set format = F22.4.
 886 DATA LIST notable LIST /X *
 887 BEGIN DATA.
 888 461.19000000
 889 466.38000000
 890 479.46000000
 891 480.10000000
 892 483.43000000
 893 488.30000000
 894 489.00000000
 895 491.62000000
 896 505.62000000
 897 511.30000000
 898 521.53000000
 899 526.70000000
 900 528.25000000
 901 538.70000000
 902 540.22000000
 903 540.58000000
 904 546.10000000
 905 548.17000000
 906 553.99000000
 907 566.21000000
 908 575.90000000
 909 584.38000000
 910 593.40000000
 911 357.05000000
 912 359.73000000
 913 360.48000000
 914 373.98000000
 915 374.13000000
 916 381.45000000
 917 383.72000000
 918 390.00000000
 919 400.34000000
 920 415.32000000
 921 415.91000000
 922 418.30000000
 923 421.03000000
 924 422.43000000
 925 426.93000000
 926 433.25000000
 927 436.89000000
 928 445.33000000
 929 446.33000000
 930 446.55000000
 931 456.44000000
 932 689.49000000
 933 691.92000000
 934 695.00000000
 935 695.36000000
 936 698.21000000
 937 699.46000000
 938 706.61000000
 939 710.69000000
 940 715.82000000
 941 715.82000000
 942 741.39000000
 943 752.27000000
 944 756.73000000
 945 757.74000000
 946 759.57000000
 947 796.07000000
 948 813.78000000
 949 817.25000000
 950 825.48000000
 951 831.28000000
 952 849.24000000
 953 890.00000000
 954 894.78000000
 955 935.65000000
 956 935.90000000
 957 945.90000000
 958 1012.8600000
 959 1022.6000000
 960 1061.8100000
 961 1063.5000000
 962 1077.2300000
 963 1151.6300000
 964 1355.2800000
 965 598.88000000
 966 606.91000000
 967 621.60000000
 968 624.80000000
 969 636.13000000
 970 637.38000000
 971 640.32000000
 972 649.35000000
 973 656.51000000
 974 662.55000000
 975 664.69000000
 976 106.22000000
 977 132.24000000
 978 174.76000000
 979 204.85000000
 980 264.93000000
 981 264.99000000
 982 269.84000000
 983 325.12000000
 984 331.67000000
 985 337.26000000
 986 347.68000000
 987 354.91000000
 988 END DATA.
 989
 990 EXAMINE
 991         x
 992         /STATISTICS=DESCRIPTIVES
 993         .
 994 ])
 995
 996 AT_CHECK([pspp -O format=csv sample.sps], [0], [dnl
 997 Table: Case Processing Summary
 998 ,Cases,,,,,
 999 ,Valid,,Missing,,Total,
1000 ,N,Percent,N,Percent,N,Percent
1001 X,100,100%,0,0%,100,100%
1002
1003 Table: Descriptives
1004 ,,,Statistic,Std. Error
1005 X,Mean,,587.6603,23.2665
1006 ,95% Confidence Interval for Mean,Lower Bound,541.4946,
1007 ,,Upper Bound,633.8260,
1008 ,5% Trimmed Mean,,579.7064,
1009 ,Median,,547.1350,
1010 ,Variance,,54132.8466,
1011 ,Std. Deviation,,232.6647,
1012 ,Minimum,,106.2200,
1013 ,Maximum,,1355.2800,
1014 ,Range,,1249.0600,
1015 ,Interquartile Range,,293.1575,
1016 ,Skewness,,.6331,.2414
1017 ,Kurtosis,,.5300,.4783
1018 ])
1019
1020 AT_CLEANUP
1021
1022
1023
1024 dnl Test for a crash which happened on bad input syntax
1025 AT_SETUP([EXAMINE -- Empty Parentheses])
1026
1027 AT_DATA([examine-empty-parens.sps], [dnl
1028 DATA LIST notable LIST /X *
1029 BEGIN DATA.
1030 2
1031 3
1032 END DATA.
1033
1034
1035 EXAMINE
1036         x
1037         /PLOT = SPREADLEVEL()
1038         .
1039 ])
1040
1041 AT_CHECK([pspp -o pspp.csv examine-empty-parens.sps], [1], [ignore])
1042
1043 AT_CLEANUP