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

   1 AT_BANNER([FREQUENCIES procedure])
   2
   3 AT_SETUP([FREQUENCIES string variable])
   4 AT_DATA([frequencies.sps],
   5   [DATA LIST FREE/
   6    name  (A8) value * quantity .
   7 BEGIN DATA.
   8 foo 1 5
   9 bar 2 6
  10 baz 1 9
  11 quux 3 1
  12 bar 1 2
  13 baz 4 3
  14 baz 1 4
  15 baz 1 1
  16 foo 6 0
  17 quux 5 8
  18 END DATA.
  19 EXECUTE.
  20
  21 FREQUENCIES /VAR = name.
  22 ])
  23 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
  24 Table: name
  25 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
  26 ,bar     ,2,20.00,20.00,20.00
  27 ,baz     ,4,40.00,40.00,60.00
  28 ,foo     ,2,20.00,20.00,80.00
  29 ,quux    ,2,20.00,20.00,100.00
  30 Total,,10,100.0,100.0,
  31 ])
  32 AT_CLEANUP
  33
  34 # Tests for a bug where pspp would crash if two FREQUENCIES commands
  35 # existed in a input file.
  36 AT_SETUP([FREQUENCIES two runs crash])
  37 AT_DATA([frequencies.sps],
  38   [data list free /v1 v2.
  39 begin data.
  40 0 1
  41 2 3
  42 4 5
  43 3 4
  44 end data.
  45
  46 frequencies v1 v2/statistics=none.
  47 frequencies v1 v2/statistics=none.
  48 ])
  49 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
  50   [Table: v1
  51 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
  52 ,.00,1,25.00,25.00,25.00
  53 ,2.00,1,25.00,25.00,50.00
  54 ,3.00,1,25.00,25.00,75.00
  55 ,4.00,1,25.00,25.00,100.00
  56 Total,,4,100.0,100.0,
  57
  58 Table: v2
  59 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
  60 ,1.00,1,25.00,25.00,25.00
  61 ,3.00,1,25.00,25.00,50.00
  62 ,4.00,1,25.00,25.00,75.00
  63 ,5.00,1,25.00,25.00,100.00
  64 Total,,4,100.0,100.0,
  65
  66 Table: v1
  67 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
  68 ,.00,1,25.00,25.00,25.00
  69 ,2.00,1,25.00,25.00,50.00
  70 ,3.00,1,25.00,25.00,75.00
  71 ,4.00,1,25.00,25.00,100.00
  72 Total,,4,100.0,100.0,
  73
  74 Table: v2
  75 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
  76 ,1.00,1,25.00,25.00,25.00
  77 ,3.00,1,25.00,25.00,50.00
  78 ,4.00,1,25.00,25.00,75.00
  79 ,5.00,1,25.00,25.00,100.00
  80 Total,,4,100.0,100.0,
  81 ])
  82 AT_CLEANUP
  83
  84 # Tests for a bug where PSPP would crash when a FREQUENCIES command
  85 # was used with the HTML output driver.
  86 AT_SETUP([FREQUENCIES HTML output crash])
  87 AT_DATA([frequencies.sps],
  88   [data list free /v1 v2.
  89 begin data.
  90 0 1
  91 2 3
  92 4 5
  93 3 4
  94 end data.
  95
  96 list.
  97
  98 frequencies v1/statistics=none.
  99 ])
 100 AT_CHECK([pspp -o - -O format=csv -o pspp.html frequencies.sps], [0],
 101   [Table: Data List
 102 v1,v2
 103 .00,1.00
 104 2.00,3.00
 105 4.00,5.00
 106 3.00,4.00
 107
 108 Table: v1
 109 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 110 ,.00,1,25.00,25.00,25.00
 111 ,2.00,1,25.00,25.00,50.00
 112 ,3.00,1,25.00,25.00,75.00
 113 ,4.00,1,25.00,25.00,100.00
 114 Total,,4,100.0,100.0,
 115 ])
 116 AT_CHECK([test -s pspp.html])
 117 AT_CLEANUP
 118
 119 # Tests for a bug which crashed PSPP when a piechart with too many
 120 # segments was requested.
 121 AT_SETUP([FREQUENCIES pie chart crash])
 122 AT_DATA([frequencies.sps],
 123   [data list list /x * w *.
 124 begin data.
 125 1  4
 126 34 10
 127 -9 15
 128 232 6
 129 11  4
 130 134 1
 131 9  5
 132 32 16
 133 -2 6
 134 2  16
 135 20  6
 136 end data.
 137
 138 weight by w.
 139
 140 frequencies /x /format=notable /statistics=none
 141         /piechart.
 142 ])
 143 # Cannot use the CSV driver for this because it does not output charts
 144 # at all.
 145 AT_CHECK([pspp frequencies.sps], [0], [dnl
 146 Reading free-form data from INLINE.
 147 +--------+------+
 148 |Variable|Format|
 149 #========#======#
 150 |x       |F8.0  |
 151 |w       |F8.0  |
 152 +--------+------+
 153 ])
 154 AT_CLEANUP
 155
 156 dnl Check that histogram subcommand runs wihout crashing
 157 AT_SETUP([FREQUENCIES histogram crash])
 158 AT_DATA([frequencies.sps],
 159   [data list notable list /x * w *.
 160 begin data.
 161 1  4
 162 34 10
 163 -9 15
 164 232 6
 165 11  4
 166 134 1
 167 9  5
 168 32 16
 169 -2 6
 170 2  16
 171 20  6
 172 end data.
 173
 174 weight by w.
 175
 176 frequencies /x
 177             /format=notable
 178             /statistics=none
 179             /histogram=minimum(0) maximum(50) percent(5) normal.
 180 ])
 181 # Cannot use the CSV driver for this because it does not output charts
 182 # at all.
 183 AT_CHECK([pspp -O format=pdf frequencies.sps], [0], [ignore])
 184 AT_CLEANUP
 185
 186 # Tests for a bug which crashed PSPP when the median and a histogram
 187 # were both requested.
 188 AT_SETUP([FREQUENCIES median with histogram crash])
 189 AT_DATA([frequencies.sps], [dnl
 190 data list list notable /x.
 191 begin data.
 192 1
 193 end data.
 194
 195 frequencies /x /histogram /STATISTICS=median.
 196 ])
 197 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [ignore])
 198 dnl Ignore output - No crash test.
 199 AT_CLEANUP
 200
 201 # Tests for a bug which caused FREQUENCIES following TEMPORARY to
 202 # crash (bug #11492).
 203 AT_SETUP([FREQUENCIES crash after TEMPORARY])
 204 AT_DATA([frequencies.sps],
 205   [DATA LIST LIST /SEX (A1) X *.
 206 BEGIN DATA.
 207 M 31
 208 F 21
 209 M 41
 210 F 31
 211 M 13
 212 F 12
 213 M 14
 214 F 13
 215 END DATA.
 216
 217
 218 TEMPORARY
 219 SELECT IF SEX EQ 'F'
 220 FREQUENCIES /X .
 221
 222 FINISH
 223 ])
 224 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 225   [Table: Reading free-form data from INLINE.
 226 Variable,Format
 227 SEX,A1
 228 X,F8.0
 229
 230 Table: X
 231 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 232 ,12.00,1,25.00,25.00,25.00
 233 ,13.00,1,25.00,25.00,50.00
 234 ,21.00,1,25.00,25.00,75.00
 235 ,31.00,1,25.00,25.00,100.00
 236 Total,,4,100.0,100.0,
 237
 238 Table: X
 239 N,Valid,4
 240 ,Missing,0
 241 Mean,,19.25
 242 Std Dev,,8.81
 243 Minimum,,12.00
 244 Maximum,,31.00
 245 ])
 246 AT_CLEANUP
 247
 248 m4_define([FREQUENCIES_NTILES_OUTPUT],
 249   [Table: x
 250 N,Valid,5
 251 ,Missing,0
 252 Mean,,3.00
 253 Std Dev,,1.58
 254 Minimum,,1.00
 255 Maximum,,5.00
 256 Percentiles,0,1.00
 257 ,25,2.00
 258 ,33,2.33
 259 ,50 (Median),3.00
 260 ,67,3.67
 261 ,75,4.00
 262 ,100,5.00
 263 ])
 264 AT_SETUP([FREQUENCIES basic percentiles])
 265 AT_DATA([frequencies.sps],
 266   [DATA LIST LIST notable /x * .
 267 BEGIN DATA.
 268 1
 269 2
 270 3
 271 4
 272 5
 273 END DATA.
 274
 275 FREQUENCIES
 276         VAR=x
 277         /FORMAT=NOTABLE
 278         /PERCENTILES = 0 25 33.333 50 66.666 75 100.
 279 ])
 280 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 281   [FREQUENCIES_NTILES_OUTPUT])
 282 AT_CLEANUP
 283
 284 AT_SETUP([FREQUENCIES basic n-tiles])
 285 AT_DATA([frequencies.sps],
 286   [DATA LIST LIST notable /x * .
 287 BEGIN DATA.
 288 1
 289 2
 290 3
 291 4
 292 5
 293 END DATA.
 294
 295 FREQUENCIES
 296         VAR=x
 297         /FORMAT=NOTABLE
 298         /NTILES = 3
 299         /NTILES = 4.
 300 ])
 301 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 302   [FREQUENCIES_NTILES_OUTPUT])
 303 AT_CLEANUP
 304
 305 AT_SETUP([FREQUENCIES compatibility percentiles])
 306 AT_DATA([frequencies.sps],
 307   [DATA LIST LIST notable /X * .
 308 BEGIN DATA.
 309 1
 310 2
 311 3
 312 4
 313 5
 314 END DATA.
 315
 316 FREQUENCIES
 317         VAR=x
 318         /ALGORITHM=COMPATIBLE
 319         /PERCENTILES = 0 25 50 75 100.
 320 ])
 321 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 322   [Table: X
 323 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 324 ,1.00,1,20.00,20.00,20.00
 325 ,2.00,1,20.00,20.00,40.00
 326 ,3.00,1,20.00,20.00,60.00
 327 ,4.00,1,20.00,20.00,80.00
 328 ,5.00,1,20.00,20.00,100.00
 329 Total,,5,100.0,100.0,
 330
 331 Table: X
 332 N,Valid,5
 333 ,Missing,0
 334 Mean,,3.00
 335 Std Dev,,1.58
 336 Minimum,,1.00
 337 Maximum,,5.00
 338 Percentiles,0,1.00
 339 ,25,1.50
 340 ,50 (Median),3.00
 341 ,75,4.50
 342 ,100,5.00
 343 ])
 344 AT_CLEANUP
 345
 346 AT_SETUP([FREQUENCIES enhanced percentiles])
 347 AT_DATA([frequencies.sps],
 348   [DATA LIST LIST notable /X * .
 349 BEGIN DATA.
 350 1
 351 2
 352 3
 353 4
 354 5
 355 END DATA.
 356
 357 FREQUENCIES
 358         VAR=x
 359         /PERCENTILES = 0 25 50 75 100.
 360 ])
 361 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 362   [Table: X
 363 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 364 ,1.00,1,20.00,20.00,20.00
 365 ,2.00,1,20.00,20.00,40.00
 366 ,3.00,1,20.00,20.00,60.00
 367 ,4.00,1,20.00,20.00,80.00
 368 ,5.00,1,20.00,20.00,100.00
 369 Total,,5,100.0,100.0,
 370
 371 Table: X
 372 N,Valid,5
 373 ,Missing,0
 374 Mean,,3.00
 375 Std Dev,,1.58
 376 Minimum,,1.00
 377 Maximum,,5.00
 378 Percentiles,0,1.00
 379 ,25,2.00
 380 ,50 (Median),3.00
 381 ,75,4.00
 382 ,100,5.00
 383 ])
 384 AT_CLEANUP
 385
 386 AT_SETUP([FREQUENCIES enhanced percentiles, weighted])
 387 AT_DATA([frequencies.sps],
 388   [DATA LIST LIST notable /X * F *.
 389 BEGIN DATA.
 390 1 2
 391 2 2
 392 3 2
 393 4 1
 394 4 1
 395 5 1
 396 5 1
 397 END DATA.
 398
 399 WEIGHT BY f.
 400
 401 FREQUENCIES
 402         VAR=x
 403         /PERCENTILES = 0 25 50 75 100.
 404 ])
 405 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 406   [Table: X
 407 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 408 ,1.00,2.00,20.00,20.00,20.00
 409 ,2.00,2.00,20.00,20.00,40.00
 410 ,3.00,2.00,20.00,20.00,60.00
 411 ,4.00,2.00,20.00,20.00,80.00
 412 ,5.00,2.00,20.00,20.00,100.00
 413 Total,,10.00,100.0,100.0,
 414
 415 Table: X
 416 N,Valid,10.00
 417 ,Missing,.00
 418 Mean,,3.00
 419 Std Dev,,1.49
 420 Minimum,,1.00
 421 Maximum,,5.00
 422 Percentiles,0,1.00
 423 ,25,2.00
 424 ,50 (Median),3.00
 425 ,75,4.00
 426 ,100,5.00
 427 ])
 428 AT_CLEANUP
 429
 430 AT_SETUP([FREQUENCIES enhanced percentiles, weighted (2)])
 431 AT_DATA([frequencies.sps],
 432   [DATA LIST LIST notable /X * F *.
 433 BEGIN DATA.
 434 1 1
 435 3 2
 436 4 1
 437 5 1
 438 5 1
 439 END DATA.
 440
 441 WEIGHT BY f.
 442
 443 FREQUENCIES
 444         VAR=x
 445         /PERCENTILES = 0 25 50 75 100.
 446 ])
 447 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 448   [Table: X
 449 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 450 ,1.00,1.00,16.67,16.67,16.67
 451 ,3.00,2.00,33.33,33.33,50.00
 452 ,4.00,1.00,16.67,16.67,66.67
 453 ,5.00,2.00,33.33,33.33,100.00
 454 Total,,6.00,100.0,100.0,
 455
 456 Table: X
 457 N,Valid,6.00
 458 ,Missing,.00
 459 Mean,,3.50
 460 Std Dev,,1.52
 461 Minimum,,1.00
 462 Maximum,,5.00
 463 Percentiles,0,1.00
 464 ,25,3.00
 465 ,50 (Median),3.50
 466 ,75,4.75
 467 ,100,5.00
 468 ])
 469 AT_CLEANUP
 470
 471 dnl Data for this test case from Fabio Bordignon <bordignon@demos.it>.
 472 AT_SETUP([FREQUENCIES enhanced percentiles, weighted (3)])
 473 AT_DATA([frequencies.sps],
 474   [DATA LIST LIST notable /X * F *.
 475 BEGIN DATA.
 476 1 7
 477 2 16
 478 3 12
 479 4 5
 480 END DATA.
 481
 482 WEIGHT BY f.
 483
 484 FREQUENCIES
 485         VAR=x
 486         /PERCENTILES = 0 25 50 75 100.
 487 ])
 488 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 489 Table: X
 490 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 491 ,1.00,7.00,17.50,17.50,17.50
 492 ,2.00,16.00,40.00,40.00,57.50
 493 ,3.00,12.00,30.00,30.00,87.50
 494 ,4.00,5.00,12.50,12.50,100.00
 495 Total,,40.00,100.0,100.0,
 496
 497 Table: X
 498 N,Valid,40.00
 499 ,Missing,.00
 500 Mean,,2.38
 501 Std Dev,,.93
 502 Minimum,,1.00
 503 Maximum,,4.00
 504 Percentiles,0,1.00
 505 ,25,2.00
 506 ,50 (Median),2.00
 507 ,75,3.00
 508 ,100,4.00
 509 ])
 510 AT_CLEANUP
 511
 512 AT_SETUP([FREQUENCIES enhanced percentiles, weighted, missing values])
 513 AT_DATA([frequencies.sps],
 514   [DATA LIST LIST notable /X * F *.
 515 BEGIN DATA.
 516 1 1
 517 3 2
 518 4 1
 519 5 1
 520 5 1
 521 99 4
 522 END DATA.
 523
 524 MISSING VALUE x (99.0) .
 525 WEIGHT BY f.
 526
 527 FREQUENCIES
 528         VAR=x
 529         /PERCENTILES = 0 25 50 75 100.
 530 ])
 531
 532 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 533   [Table: X
 534 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 535 ,1.00,1.00,10.00,16.67,16.67
 536 ,3.00,2.00,20.00,33.33,50.00
 537 ,4.00,1.00,10.00,16.67,66.67
 538 ,5.00,2.00,20.00,33.33,100.00
 539 ,99.00,4.00,40.00,Missing,
 540 Total,,10.00,100.0,100.0,
 541
 542 Table: X
 543 N,Valid,6.00
 544 ,Missing,4.00
 545 Mean,,3.50
 546 Std Dev,,1.52
 547 Minimum,,1.00
 548 Maximum,,5.00
 549 Percentiles,0,1.00
 550 ,25,3.00
 551 ,50 (Median),3.50
 552 ,75,4.75
 553 ,100,5.00
 554 ])
 555 AT_CLEANUP
 556
 557 AT_SETUP([FREQUENCIES dichotomous histogram])
 558 AT_DATA([frequencies.sps], [dnl
 559 data list notable list /d4 *.
 560 begin data.
 561 0
 562 0
 563 0
 564 1
 565 0
 566 0
 567 0
 568 0
 569 1
 570 0
 571 0
 572 0
 573 0
 574 0
 575 1
 576 2
 577 0
 578 end data.
 579
 580 FREQUENCIES
 581         /VARIABLES = d4
 582         /FORMAT=AVALUE TABLE
 583         /HISTOGRAM=NORMAL
 584         .
 585 ])
 586
 587 AT_CHECK([pspp frequencies.sps], [0],  [ignore])
 588 AT_CLEANUP
 589
 590
 591 AT_SETUP([FREQUENCIES median])
 592 AT_DATA([median.sps], [dnl
 593 data list notable list /x *.
 594 begin data.
 595 1
 596 2
 597 3000000
 598 end data.
 599
 600 FREQUENCIES
 601         /VARIABLES = x
 602         /STATISTICS = MEDIAN
 603         .
 604 ])
 605
 606 AT_CHECK([pspp median.sps -O format=csv], [0],  [dnl
 607 Table: x
 608 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 609 ,1.00,1,33.33,33.33,33.33
 610 ,2.00,1,33.33,33.33,66.67
 611 ,3000000.00,1,33.33,33.33,100.00
 612 Total,,3,100.0,100.0,
 613
 614 Table: x
 615 N,Valid,3
 616 ,Missing,0
 617 Percentiles,50 (Median),2.00
 618 ])
 619 AT_CLEANUP
 620
 621
 622
 623 AT_SETUP([FREQUENCIES default statistics])
 624 AT_DATA([median.sps], [dnl
 625 data list notable list /x *.
 626 begin data.
 627 10
 628 20
 629 3000000
 630 end data.
 631
 632 FREQUENCIES
 633         /VARIABLES = x
 634         /STATISTICS
 635         .
 636
 637 FREQUENCIES
 638         /VARIABLES = x
 639         /STATISTICS = DEFAULT
 640         .
 641 ])
 642
 643 AT_CHECK([pspp median.sps -O format=csv], [0],  [dnl
 644 Table: x
 645 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 646 ,10.00,1,33.33,33.33,33.33
 647 ,20.00,1,33.33,33.33,66.67
 648 ,3000000.00,1,33.33,33.33,100.00
 649 Total,,3,100.0,100.0,
 650
 651 Table: x
 652 N,Valid,3
 653 ,Missing,0
 654 Mean,,1000010.00
 655 Std Dev,,1732042.15
 656 Minimum,,10.00
 657 Maximum,,3000000.00
 658
 659 Table: x
 660 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 661 ,10.00,1,33.33,33.33,33.33
 662 ,20.00,1,33.33,33.33,66.67
 663 ,3000000.00,1,33.33,33.33,100.00
 664 Total,,3,100.0,100.0,
 665
 666 Table: x
 667 N,Valid,3
 668 ,Missing,0
 669 Mean,,1000010.00
 670 Std Dev,,1732042.15
 671 Minimum,,10.00
 672 Maximum,,3000000.00
 673 ])
 674 AT_CLEANUP
 675
 676
 677
 678 AT_SETUP([FREQUENCIES no valid data])
 679 AT_DATA([empty.sps], [dnl
 680 data list notable list /x *.
 681 begin data.
 682 .
 683 .
 684 .
 685 end data.
 686
 687 FREQUENCIES
 688         /VARIABLES = x
 689         /STATISTICS = ALL
 690         .
 691 ])
 692
 693 AT_CHECK([pspp empty.sps -O format=csv], [0],  [dnl
 694 Table: x
 695 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 696 ,.  ,3,100.00,Missing,
 697 Total,,3,100.0,100.0,
 698
 699 Table: x
 700 N,Valid,0
 701 ,Missing,3
 702 Mean,,.
 703 S.E. Mean,,.
 704 Mode,,.
 705 Std Dev,,.
 706 Variance,,.
 707 Kurtosis,,.
 708 S.E. Kurt,,.
 709 Skewness,,.
 710 S.E. Skew,,.
 711 Range,,.
 712 Minimum,,.
 713 Maximum,,.
 714 Sum,,.
 715 Percentiles,,.
 716 ])
 717
 718 AT_CLEANUP