pintos-os.org Git - pspp/blob - tests/language/stats/frequencies.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([FREQUENCIES procedure])
  18
  19 AT_SETUP([FREQUENCIES string variable])
  20 AT_DATA([frequencies.sps],
  21   [DATA LIST FREE/
  22    name  (A8) value * quantity .
  23 BEGIN DATA.
  24 foo 1 5
  25 bar 2 6
  26 baz 1 9
  27 quux 3 1
  28 bar 1 2
  29 baz 4 3
  30 baz 1 4
  31 baz 1 1
  32 foo 6 0
  33 quux 5 8
  34 END DATA.
  35 EXECUTE.
  36
  37 FREQUENCIES /VAR = name/ORDER=ANALYSIS.
  38 ])
  39 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
  40 Table: name
  41 ,,Frequency,Percent,Valid Percent,Cumulative Percent
  42 Valid,bar,2,20.0%,20.0%,20.0%
  43 ,baz,4,40.0%,40.0%,60.0%
  44 ,foo,2,20.0%,20.0%,80.0%
  45 ,quux,2,20.0%,20.0%,100.0%
  46 Total,,10,100.0%,,
  47 ])
  48 AT_CLEANUP
  49
  50 AT_SETUP([FREQUENCIES with SPLIT FILE - LAYERED])
  51 AT_DATA([frequencies.sps], [dnl
  52 DATA LIST LIST NOTABLE/name (A8) value quantity.
  53 BEGIN DATA.
  54 foo 1 5
  55 bar 2 6
  56 baz 1 9
  57 quux 3 1
  58 bar 1 2
  59 baz 4 3
  60 baz 1 4
  61 baz 1 1
  62 foo 6 0
  63 quux 5 8
  64 END DATA.
  65 EXECUTE.
  66
  67 SORT CASES BY name.
  68 SPLIT FILE BY name.
  69 FREQUENCIES /VARIABLES=value quantity /FORMAT NOTABLE.
  70 ])
  71 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
  72 Table: Statistics
  73 name,,,value,quantity
  74 bar,N,Valid,2,2
  75 ,,Missing,0,0
  76 ,Mean,,1.50,4.00
  77 ,Std Dev,,.71,2.83
  78 ,Minimum,,1.00,2.00
  79 ,Maximum,,2.00,6.00
  80 baz,N,Valid,4,4
  81 ,,Missing,0,0
  82 ,Mean,,1.75,4.25
  83 ,Std Dev,,1.50,3.40
  84 ,Minimum,,1.00,1.00
  85 ,Maximum,,4.00,9.00
  86 foo,N,Valid,2,2
  87 ,,Missing,0,0
  88 ,Mean,,3.50,2.50
  89 ,Std Dev,,3.54,3.54
  90 ,Minimum,,1.00,.00
  91 ,Maximum,,6.00,5.00
  92 quux,N,Valid,2,2
  93 ,,Missing,0,0
  94 ,Mean,,4.00,4.50
  95 ,Std Dev,,1.41,4.95
  96 ,Minimum,,3.00,1.00
  97 ,Maximum,,5.00,8.00
  98 ])
  99 AT_CLEANUP
 100
 101 AT_SETUP([FREQUENCIES with SPLIT FILE - SEPARATE])
 102 AT_DATA([frequencies.sps], [dnl
 103 DATA LIST LIST NOTABLE/name (A8) value quantity.
 104 BEGIN DATA.
 105 foo 1 5
 106 bar 2 6
 107 baz 1 9
 108 quux 3 1
 109 bar 1 2
 110 baz 4 3
 111 baz 1 4
 112 baz 1 1
 113 foo 6 0
 114 quux 5 8
 115 END DATA.
 116 EXECUTE.
 117
 118 SORT CASES BY name.
 119 SPLIT FILE SEPARATE BY name.
 120 FREQUENCIES /VARIABLES=value quantity /FORMAT NOTABLE.
 121 ])
 122 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 123 Table: Split Values
 124 Variable,Value
 125 name,bar
 126
 127 Table: Statistics
 128 ,,value,quantity
 129 N,Valid,2,2
 130 ,Missing,0,0
 131 Mean,,1.50,4.00
 132 Std Dev,,.71,2.83
 133 Minimum,,1.00,2.00
 134 Maximum,,2.00,6.00
 135
 136 Table: Split Values
 137 Variable,Value
 138 name,baz
 139
 140 Table: Statistics
 141 ,,value,quantity
 142 N,Valid,4,4
 143 ,Missing,0,0
 144 Mean,,1.75,4.25
 145 Std Dev,,1.50,3.40
 146 Minimum,,1.00,1.00
 147 Maximum,,4.00,9.00
 148
 149 Table: Split Values
 150 Variable,Value
 151 name,foo
 152
 153 Table: Statistics
 154 ,,value,quantity
 155 N,Valid,2,2
 156 ,Missing,0,0
 157 Mean,,3.50,2.50
 158 Std Dev,,3.54,3.54
 159 Minimum,,1.00,.00
 160 Maximum,,6.00,5.00
 161
 162 Table: Split Values
 163 Variable,Value
 164 name,quux
 165
 166 Table: Statistics
 167 ,,value,quantity
 168 N,Valid,2,2
 169 ,Missing,0,0
 170 Mean,,4.00,4.50
 171 Std Dev,,1.41,4.95
 172 Minimum,,3.00,1.00
 173 Maximum,,5.00,8.00
 174 ])
 175 AT_CLEANUP
 176
 177 AT_SETUP([FREQUENCIES with SPLIT FILE - LAYERED - unsorted data])
 178 AT_DATA([frequencies.sps], [dnl
 179 DATA LIST LIST NOTABLE/name (A8) value quantity.
 180 BEGIN DATA.
 181 foo 1 5
 182 bar 2 6
 183 baz 1 9
 184 quux 3 1
 185 baz 4 3
 186 bar 1 2
 187 baz 1 1
 188 foo 6 0
 189 baz 1 4
 190 quux 5 8
 191 END DATA.
 192 EXECUTE.
 193
 194 SPLIT FILE BY name.
 195 FREQUENCIES /VARIABLES=value quantity /FORMAT NOTABLE.
 196 ])
 197 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 198 "frequencies.sps:17: warning: FREQUENCIES: When SPLIT FILE is in effect, the input data must be sorted by the split variables (for example, using SORT CASES), but multiple runs of cases with the same split values were found separated by cases with different values.  Each run will be analyzed separately.  The duplicate split values are: name = baz     "
 199
 200 "frequencies.sps:17: warning: FREQUENCIES: When SPLIT FILE is in effect, the input data must be sorted by the split variables (for example, using SORT CASES), but multiple runs of cases with the same split values were found separated by cases with different values.  Each run will be analyzed separately.  The duplicate split values are: name = bar     "
 201
 202 "frequencies.sps:17: warning: FREQUENCIES: When SPLIT FILE is in effect, the input data must be sorted by the split variables (for example, using SORT CASES), but multiple runs of cases with the same split values were found separated by cases with different values.  Each run will be analyzed separately.  The duplicate split values are: name = baz     "
 203
 204 "frequencies.sps:17: warning: FREQUENCIES: When SPLIT FILE is in effect, the input data must be sorted by the split variables (for example, using SORT CASES), but multiple runs of cases with the same split values were found separated by cases with different values.  Each run will be analyzed separately.  The duplicate split values are: name = foo     "
 205
 206 "frequencies.sps:17: warning: FREQUENCIES: When SPLIT FILE is in effect, the input data must be sorted by the split variables (for example, using SORT CASES), but multiple runs of cases with the same split values were found separated by cases with different values.  Each run will be analyzed separately.  The duplicate split values are: name = baz     "
 207
 208 Table: Statistics
 209 name,,,value,quantity
 210 foo,N,Valid,1,1
 211 ,,Missing,0,0
 212 ,Mean,,1.00,5.00
 213 ,Std Dev,,NaN,NaN
 214 ,Minimum,,1.00,5.00
 215 ,Maximum,,1.00,5.00
 216 bar,N,Valid,1,1
 217 ,,Missing,0,0
 218 ,Mean,,2.00,6.00
 219 ,Std Dev,,NaN,NaN
 220 ,Minimum,,2.00,6.00
 221 ,Maximum,,2.00,6.00
 222 baz,N,Valid,1,1
 223 ,,Missing,0,0
 224 ,Mean,,1.00,9.00
 225 ,Std Dev,,NaN,NaN
 226 ,Minimum,,1.00,9.00
 227 ,Maximum,,1.00,9.00
 228 quux,N,Valid,1,1
 229 ,,Missing,0,0
 230 ,Mean,,3.00,1.00
 231 ,Std Dev,,NaN,NaN
 232 ,Minimum,,3.00,1.00
 233 ,Maximum,,3.00,1.00
 234 baz,N,Valid,1,1
 235 ,,Missing,0,0
 236 ,Mean,,4.00,3.00
 237 ,Std Dev,,NaN,NaN
 238 ,Minimum,,4.00,3.00
 239 ,Maximum,,4.00,3.00
 240 bar,N,Valid,1,1
 241 ,,Missing,0,0
 242 ,Mean,,1.00,2.00
 243 ,Std Dev,,NaN,NaN
 244 ,Minimum,,1.00,2.00
 245 ,Maximum,,1.00,2.00
 246 baz,N,Valid,1,1
 247 ,,Missing,0,0
 248 ,Mean,,1.00,1.00
 249 ,Std Dev,,NaN,NaN
 250 ,Minimum,,1.00,1.00
 251 ,Maximum,,1.00,1.00
 252 foo,N,Valid,1,1
 253 ,,Missing,0,0
 254 ,Mean,,6.00,.00
 255 ,Std Dev,,NaN,NaN
 256 ,Minimum,,6.00,.00
 257 ,Maximum,,6.00,.00
 258 baz,N,Valid,1,1
 259 ,,Missing,0,0
 260 ,Mean,,1.00,4.00
 261 ,Std Dev,,NaN,NaN
 262 ,Minimum,,1.00,4.00
 263 ,Maximum,,1.00,4.00
 264 quux,N,Valid,1,1
 265 ,,Missing,0,0
 266 ,Mean,,5.00,8.00
 267 ,Std Dev,,NaN,NaN
 268 ,Minimum,,5.00,8.00
 269 ,Maximum,,5.00,8.00
 270
 271 frequencies.sps:17: warning: FREQUENCIES: Suppressed 1 additional warning about duplicate split values.
 272 ])
 273 AT_CLEANUP
 274
 275 # Tests for a bug where pspp would crash if two FREQUENCIES commands
 276 # existed in a input file.
 277 AT_SETUP([FREQUENCIES two runs crash])
 278 AT_DATA([frequencies.sps],
 279   [data list free /v1 v2.
 280 begin data.
 281 0 1
 282 2 3
 283 4 5
 284 3 4
 285 end data.
 286
 287 frequencies v1 v2/statistics=none/ORDER=VARIABLE.
 288 frequencies v1 v2/statistics=none.
 289 ])
 290 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 291 Table: v1
 292 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 293 Valid,.00,1,25.0%,25.0%,25.0%
 294 ,2.00,1,25.0%,25.0%,50.0%
 295 ,3.00,1,25.0%,25.0%,75.0%
 296 ,4.00,1,25.0%,25.0%,100.0%
 297 Total,,4,100.0%,,
 298
 299 Table: v2
 300 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 301 Valid,1.00,1,25.0%,25.0%,25.0%
 302 ,3.00,1,25.0%,25.0%,50.0%
 303 ,4.00,1,25.0%,25.0%,75.0%
 304 ,5.00,1,25.0%,25.0%,100.0%
 305 Total,,4,100.0%,,
 306
 307 Table: v1
 308 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 309 Valid,.00,1,25.0%,25.0%,25.0%
 310 ,2.00,1,25.0%,25.0%,50.0%
 311 ,3.00,1,25.0%,25.0%,75.0%
 312 ,4.00,1,25.0%,25.0%,100.0%
 313 Total,,4,100.0%,,
 314
 315 Table: v2
 316 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 317 Valid,1.00,1,25.0%,25.0%,25.0%
 318 ,3.00,1,25.0%,25.0%,50.0%
 319 ,4.00,1,25.0%,25.0%,75.0%
 320 ,5.00,1,25.0%,25.0%,100.0%
 321 Total,,4,100.0%,,
 322 ])
 323 AT_CLEANUP
 324
 325 # Test that the LIMIT specification works.
 326 AT_SETUP([FREQUENCIES with LIMIT])
 327 AT_DATA([frequencies.sps],
 328   [data list free /v1 v2.
 329 begin data.
 330 0 1
 331 2 5
 332 4 3
 333 3 5
 334 end data.
 335
 336 frequencies v1 v2/statistics=none/FORMAT=LIMIT(3).
 337 ])
 338 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 339 Table: v2
 340 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 341 Valid,1.00,1,25.0%,25.0%,25.0%
 342 ,3.00,1,25.0%,25.0%,50.0%
 343 ,5.00,2,50.0%,50.0%,100.0%
 344 Total,,4,100.0%,,
 345 ])
 346 AT_CLEANUP
 347
 348 # Tests for a bug where PSPP would crash when a FREQUENCIES command
 349 # was used with the HTML output driver.
 350 AT_SETUP([FREQUENCIES HTML output crash])
 351 AT_DATA([frequencies.sps],
 352   [data list free /v1 v2.
 353 begin data.
 354 0 1
 355 2 3
 356 4 5
 357 3 4
 358 end data.
 359
 360 list.
 361
 362 frequencies v1/statistics=none.
 363 ])
 364 AT_CHECK([pspp -o - -O format=csv -o pspp.html frequencies.sps], [0],
 365   [Table: Data List
 366 v1,v2
 367 .00,1.00
 368 2.00,3.00
 369 4.00,5.00
 370 3.00,4.00
 371
 372 Table: v1
 373 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 374 Valid,.00,1,25.0%,25.0%,25.0%
 375 ,2.00,1,25.0%,25.0%,50.0%
 376 ,3.00,1,25.0%,25.0%,75.0%
 377 ,4.00,1,25.0%,25.0%,100.0%
 378 Total,,4,100.0%,,
 379 ])
 380 AT_CHECK([test -s pspp.html])
 381 AT_CLEANUP
 382
 383 # Tests for a bug which crashed PSPP when a piechart with too many
 384 # segments was requested.
 385 AT_SETUP([FREQUENCIES pie chart crash])
 386 AT_DATA([frequencies.sps],
 387   [data list list /x * w *.
 388 begin data.
 389 1  4
 390 34 10
 391 -9 15
 392 232 6
 393 11  4
 394 134 1
 395 9  5
 396 32 16
 397 -2 6
 398 2  16
 399 20  6
 400 end data.
 401
 402 weight by w.
 403
 404 frequencies /x /format=notable /statistics=none
 405         /piechart.
 406 ])
 407 # Cannot use the CSV driver for this because it does not output charts
 408 # at all.
 409 AT_CHECK([pspp frequencies.sps], [0], [dnl
 410 Reading free-form data from INLINE.
 411 +--------+------+
 412 |Variable|Format|
 413 +--------+------+
 414 |x       |F8.0  |
 415 |w       |F8.0  |
 416 +--------+------+
 417 ])
 418 AT_CLEANUP
 419
 420 dnl Check that histogram subcommand runs wihout crashing
 421 AT_SETUP([FREQUENCIES histogram crash])
 422 AT_DATA([frequencies.sps],
 423   [data list notable list /x * w *.
 424 begin data.
 425 1  4
 426 34 10
 427 -9 15
 428 232 6
 429 11  4
 430 134 1
 431 9  5
 432 32 16
 433 -2 6
 434 2  16
 435 20  6
 436 end data.
 437
 438 weight by w.
 439
 440 frequencies /x
 441             /format=notable
 442             /statistics=none
 443             /histogram=minimum(0) maximum(50) percent(5) normal.
 444 ])
 445 # Cannot use the CSV driver for this because it does not output charts
 446 # at all.
 447 AT_CHECK([pspp -O format=pdf frequencies.sps], [0], [ignore], [ignore])
 448 AT_CLEANUP
 449
 450 # Tests for a bug which crashed PSPP when the median and a histogram
 451 # were both requested.
 452 AT_SETUP([FREQUENCIES median with histogram crash])
 453 AT_DATA([frequencies.sps], [dnl
 454 data list list notable /x.
 455 begin data.
 456 1
 457 end data.
 458
 459 frequencies /x /histogram /STATISTICS=median.
 460 ])
 461 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [ignore])
 462 dnl Ignore output - No crash test.
 463 AT_CLEANUP
 464
 465 # Tests for a bug which caused FREQUENCIES following TEMPORARY to
 466 # crash (bug #11492).
 467 AT_SETUP([FREQUENCIES crash after TEMPORARY])
 468 AT_DATA([frequencies.sps],
 469   [DATA LIST LIST /SEX (A1) X *.
 470 BEGIN DATA.
 471 M 31
 472 F 21
 473 M 41
 474 F 31
 475 M 13
 476 F 12
 477 M 14
 478 F 13
 479 END DATA.
 480
 481
 482 TEMPORARY
 483 SELECT IF SEX EQ 'F'
 484 FREQUENCIES /X .
 485
 486 FINISH
 487 ])
 488 AT_CHECK([pspp -o pspp.csv -o pspp.txt frequencies.sps])
 489 AT_CHECK([cat pspp.csv], [0], [dnl
 490 Table: Reading free-form data from INLINE.
 491 Variable,Format
 492 SEX,A1
 493 X,F8.0
 494
 495 Table: Statistics
 496 ,,X
 497 N,Valid,4
 498 ,Missing,0
 499 Mean,,19.25
 500 Std Dev,,8.81
 501 Minimum,,12.00
 502 Maximum,,31.00
 503
 504 Table: X
 505 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 506 Valid,12.00,1,25.0%,25.0%,25.0%
 507 ,13.00,1,25.0%,25.0%,50.0%
 508 ,21.00,1,25.0%,25.0%,75.0%
 509 ,31.00,1,25.0%,25.0%,100.0%
 510 Total,,4,100.0%,,
 511 ])
 512 AT_CLEANUP
 513
 514 m4_define([FREQUENCIES_NTILES_OUTPUT], [dnl
 515 Table: Statistics
 516 ,,x,y
 517 N,Valid,5,5
 518 ,Missing,0,0
 519 Mean,,3.00,30.00
 520 Std Dev,,1.58,15.81
 521 Minimum,,1.00,10.00
 522 Maximum,,5.00,50.00
 523 Percentiles,0,1.00,10.00
 524 ,25,2.00,20.00
 525 ,33,2.33,23.33
 526 ,50,3.00,30.00
 527 ,67,3.67,36.67
 528 ,75,4.00,40.00
 529 ,100,5.00,50.00
 530 ])
 531 AT_SETUP([FREQUENCIES basic percentiles])
 532 AT_DATA([frequencies.sps],
 533   [DATA LIST LIST notable /x y.
 534 BEGIN DATA.
 535 1 10
 536 2 20
 537 3 30
 538 4 40
 539 5 50
 540 END DATA.
 541
 542 FREQUENCIES
 543         VAR=x y
 544         /FORMAT=NOTABLE
 545         /PERCENTILES = 0 25 33.333 50 66.666 75 100.
 546 ])
 547 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 548   [FREQUENCIES_NTILES_OUTPUT])
 549 AT_CLEANUP
 550
 551 AT_SETUP([FREQUENCIES basic n-tiles])
 552 AT_DATA([frequencies.sps],
 553   [DATA LIST LIST notable /x y.
 554 BEGIN DATA.
 555 1 10
 556 2 20
 557 3 30
 558 4 40
 559 5 50
 560 END DATA.
 561
 562 FREQUENCIES
 563         VAR=x y
 564         /FORMAT=NOTABLE
 565         /NTILES = 3
 566         /NTILES = 4.
 567 ])
 568 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 569   [FREQUENCIES_NTILES_OUTPUT])
 570 AT_CLEANUP
 571
 572 AT_SETUP([FREQUENCIES compatibility percentiles])
 573 AT_DATA([frequencies.sps],
 574   [DATA LIST LIST notable /X * .
 575 BEGIN DATA.
 576 1
 577 2
 578 3
 579 4
 580 5
 581 END DATA.
 582
 583 FREQUENCIES
 584         VAR=x
 585         /ALGORITHM=COMPATIBLE
 586         /PERCENTILES = 0 25 50 75 100.
 587 ])
 588 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 589 Table: Statistics
 590 ,,X
 591 N,Valid,5
 592 ,Missing,0
 593 Mean,,3.00
 594 Std Dev,,1.58
 595 Minimum,,1.00
 596 Maximum,,5.00
 597 Percentiles,0,1.00
 598 ,25,1.50
 599 ,50,3.00
 600 ,75,4.50
 601 ,100,5.00
 602
 603 Table: X
 604 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 605 Valid,1.00,1,20.0%,20.0%,20.0%
 606 ,2.00,1,20.0%,20.0%,40.0%
 607 ,3.00,1,20.0%,20.0%,60.0%
 608 ,4.00,1,20.0%,20.0%,80.0%
 609 ,5.00,1,20.0%,20.0%,100.0%
 610 Total,,5,100.0%,,
 611 ])
 612 AT_CLEANUP
 613
 614 AT_SETUP([FREQUENCIES enhanced percentiles])
 615 AT_DATA([frequencies.sps],
 616   [DATA LIST LIST notable /X * .
 617 BEGIN DATA.
 618 1
 619 2
 620 3
 621 4
 622 5
 623 END DATA.
 624
 625 FREQUENCIES
 626         VAR=x
 627         /PERCENTILES = 0 25 50 75 100.
 628 ])
 629 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 630 Table: Statistics
 631 ,,X
 632 N,Valid,5
 633 ,Missing,0
 634 Mean,,3.00
 635 Std Dev,,1.58
 636 Minimum,,1.00
 637 Maximum,,5.00
 638 Percentiles,0,1.00
 639 ,25,2.00
 640 ,50,3.00
 641 ,75,4.00
 642 ,100,5.00
 643
 644 Table: X
 645 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 646 Valid,1.00,1,20.0%,20.0%,20.0%
 647 ,2.00,1,20.0%,20.0%,40.0%
 648 ,3.00,1,20.0%,20.0%,60.0%
 649 ,4.00,1,20.0%,20.0%,80.0%
 650 ,5.00,1,20.0%,20.0%,100.0%
 651 Total,,5,100.0%,,
 652 ])
 653 AT_CLEANUP
 654
 655 AT_SETUP([FREQUENCIES enhanced percentiles, weighted])
 656 AT_DATA([frequencies.sps],
 657   [DATA LIST LIST notable /X * F *.
 658 BEGIN DATA.
 659 1 2
 660 2 2
 661 3 2
 662 4 1
 663 4 1
 664 5 1
 665 5 1
 666 END DATA.
 667
 668 WEIGHT BY f.
 669
 670 FREQUENCIES
 671         VAR=x
 672         /PERCENTILES = 0 25 50 75 100.
 673 ])
 674 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 675 Table: Statistics
 676 ,,X
 677 N,Valid,10.00
 678 ,Missing,.00
 679 Mean,,3.00
 680 Std Dev,,1.49
 681 Minimum,,1.00
 682 Maximum,,5.00
 683 Percentiles,0,1.00
 684 ,25,2.00
 685 ,50,3.00
 686 ,75,4.00
 687 ,100,5.00
 688
 689 Table: X
 690 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 691 Valid,1.00,2.00,20.0%,20.0%,20.0%
 692 ,2.00,2.00,20.0%,20.0%,40.0%
 693 ,3.00,2.00,20.0%,20.0%,60.0%
 694 ,4.00,2.00,20.0%,20.0%,80.0%
 695 ,5.00,2.00,20.0%,20.0%,100.0%
 696 Total,,10.00,100.0%,,
 697 ])
 698 AT_CLEANUP
 699
 700 AT_SETUP([FREQUENCIES enhanced percentiles, weighted (2)])
 701 AT_DATA([frequencies.sps],
 702   [DATA LIST LIST notable /X * F *.
 703 BEGIN DATA.
 704 1 1
 705 3 2
 706 4 1
 707 5 1
 708 5 1
 709 END DATA.
 710
 711 WEIGHT BY f.
 712
 713 FREQUENCIES
 714         VAR=x
 715         /PERCENTILES = 0 25 50 75 100.
 716 ])
 717 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 718 Table: Statistics
 719 ,,X
 720 N,Valid,6.00
 721 ,Missing,.00
 722 Mean,,3.50
 723 Std Dev,,1.52
 724 Minimum,,1.00
 725 Maximum,,5.00
 726 Percentiles,0,1.00
 727 ,25,3.00
 728 ,50,3.50
 729 ,75,4.75
 730 ,100,5.00
 731
 732 Table: X
 733 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 734 Valid,1.00,1.00,16.7%,16.7%,16.7%
 735 ,3.00,2.00,33.3%,33.3%,50.0%
 736 ,4.00,1.00,16.7%,16.7%,66.7%
 737 ,5.00,2.00,33.3%,33.3%,100.0%
 738 Total,,6.00,100.0%,,
 739 ])
 740 AT_CLEANUP
 741
 742 dnl Data for this test case from Fabio Bordignon <bordignon@demos.it>.
 743 AT_SETUP([FREQUENCIES enhanced percentiles, weighted (3)])
 744 AT_DATA([frequencies.sps],
 745   [DATA LIST LIST notable /X * F *.
 746 BEGIN DATA.
 747 1 7
 748 2 16
 749 3 12
 750 4 5
 751 END DATA.
 752
 753 WEIGHT BY f.
 754
 755 FREQUENCIES
 756         VAR=x
 757         /PERCENTILES = 0 25 50 75 100.
 758 ])
 759 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 760 Table: Statistics
 761 ,,X
 762 N,Valid,40.00
 763 ,Missing,.00
 764 Mean,,2.38
 765 Std Dev,,.93
 766 Minimum,,1.00
 767 Maximum,,4.00
 768 Percentiles,0,1.00
 769 ,25,2.00
 770 ,50,2.00
 771 ,75,3.00
 772 ,100,4.00
 773
 774 Table: X
 775 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 776 Valid,1.00,7.00,17.5%,17.5%,17.5%
 777 ,2.00,16.00,40.0%,40.0%,57.5%
 778 ,3.00,12.00,30.0%,30.0%,87.5%
 779 ,4.00,5.00,12.5%,12.5%,100.0%
 780 Total,,40.00,100.0%,,
 781 ])
 782 AT_CLEANUP
 783
 784 AT_SETUP([FREQUENCIES enhanced percentiles, weighted, missing values])
 785 AT_DATA([frequencies.sps],
 786   [DATA LIST LIST notable /X * F *.
 787 BEGIN DATA.
 788 1 1
 789 3 2
 790 4 1
 791 5 1
 792 5 1
 793 99 4
 794 END DATA.
 795
 796 MISSING VALUE x (99.0) .
 797 WEIGHT BY f.
 798
 799 FREQUENCIES
 800         VAR=x
 801         /PERCENTILES = 0 25 50 75 100.
 802 ])
 803
 804 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 805 Table: Statistics
 806 ,,X
 807 N,Valid,6.00
 808 ,Missing,4.00
 809 Mean,,3.50
 810 Std Dev,,1.52
 811 Minimum,,1.00
 812 Maximum,,5.00
 813 Percentiles,0,1.00
 814 ,25,3.00
 815 ,50,3.50
 816 ,75,4.75
 817 ,100,5.00
 818
 819 Table: X
 820 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 821 Valid,1.00,1.00,10.0%,16.7%,16.7%
 822 ,3.00,2.00,20.0%,33.3%,50.0%
 823 ,4.00,1.00,10.0%,16.7%,66.7%
 824 ,5.00,2.00,20.0%,33.3%,100.0%
 825 Missing,99.00,4.00,40.0%,,
 826 Total,,10.00,100.0%,,
 827 ])
 828 AT_CLEANUP
 829
 830 AT_SETUP([FREQUENCIES dichotomous histogram])
 831 AT_DATA([frequencies.sps], [dnl
 832 data list notable list /d4 *.
 833 begin data.
 834 0
 835 0
 836 0
 837 1
 838 0
 839 0
 840 0
 841 0
 842 1
 843 0
 844 0
 845 0
 846 0
 847 0
 848 1
 849 2
 850 0
 851 end data.
 852
 853 FREQUENCIES
 854         /VARIABLES = d4
 855         /FORMAT=AVALUE TABLE
 856         /HISTOGRAM=NORMAL
 857         .
 858 ])
 859
 860 AT_CHECK([pspp frequencies.sps], [0],  [ignore])
 861 AT_CLEANUP
 862
 863
 864 AT_SETUP([FREQUENCIES median])
 865 AT_DATA([median.sps], [dnl
 866 data list notable list /x *.
 867 begin data.
 868 1
 869 2
 870 3000000
 871 end data.
 872
 873 FREQUENCIES
 874         /VARIABLES = x
 875         /STATISTICS = MEDIAN
 876         .
 877 ])
 878
 879 AT_CHECK([pspp median.sps -O format=csv], [0], [dnl
 880 Table: Statistics
 881 ,,x
 882 N,Valid,3
 883 ,Missing,0
 884 Median,,2.00
 885
 886 Table: x
 887 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 888 Valid,1.00,1,33.3%,33.3%,33.3%
 889 ,2.00,1,33.3%,33.3%,66.7%
 890 ,3000000,1,33.3%,33.3%,100.0%
 891 Total,,3,100.0%,,
 892 ])
 893 AT_CLEANUP
 894
 895 AT_SETUP([FREQUENCIES variance])
 896 AT_DATA([variance.sps], [dnl
 897 data list notable list /forename (A12) height.
 898 begin data.
 899 Ahmed 188
 900 bertram 167
 901 Catherine 134
 902 David 109
 903 end data.
 904
 905 FREQUENCIES
 906    /VARIABLES = height
 907    /STATISTICS = VARIANCE.
 908 ])
 909
 910 AT_CHECK([pspp variance.sps -O format=csv], [0], [dnl
 911 Table: Statistics
 912 ,,height
 913 N,Valid,4
 914 ,Missing,0
 915 Variance,,1223.00
 916
 917 Table: height
 918 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 919 Valid,109.00,1,25.0%,25.0%,25.0%
 920 ,134.00,1,25.0%,25.0%,50.0%
 921 ,167.00,1,25.0%,25.0%,75.0%
 922 ,188.00,1,25.0%,25.0%,100.0%
 923 Total,,4,100.0%,,
 924 ])
 925 AT_CLEANUP
 926
 927 AT_SETUP([FREQUENCIES default statistics])
 928 AT_DATA([median.sps], [dnl
 929 data list notable list /x *.
 930 begin data.
 931 10
 932 20
 933 3000000
 934 end data.
 935
 936 FREQUENCIES
 937         /VARIABLES = x
 938         /STATISTICS
 939         .
 940
 941 FREQUENCIES
 942         /VARIABLES = x
 943         /STATISTICS = DEFAULT
 944         .
 945 ])
 946
 947 AT_CHECK([pspp median.sps -o pspp.csv -o pspp.txt])
 948 AT_CHECK([cat pspp.csv], [0], [dnl
 949 Table: Statistics
 950 ,,x
 951 N,Valid,3
 952 ,Missing,0
 953 Mean,,1000010
 954 Std Dev,,1732042
 955 Minimum,,10.00
 956 Maximum,,3000000
 957
 958 Table: x
 959 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 960 Valid,10.00,1,33.3%,33.3%,33.3%
 961 ,20.00,1,33.3%,33.3%,66.7%
 962 ,3000000,1,33.3%,33.3%,100.0%
 963 Total,,3,100.0%,,
 964
 965 Table: Statistics
 966 ,,x
 967 N,Valid,3
 968 ,Missing,0
 969 Mean,,1000010
 970 Std Dev,,1732042
 971 Minimum,,10.00
 972 Maximum,,3000000
 973
 974 Table: x
 975 ,,Frequency,Percent,Valid Percent,Cumulative Percent
 976 Valid,10.00,1,33.3%,33.3%,33.3%
 977 ,20.00,1,33.3%,33.3%,66.7%
 978 ,3000000,1,33.3%,33.3%,100.0%
 979 Total,,3,100.0%,,
 980 ])
 981 AT_CLEANUP
 982
 983
 984
 985 AT_SETUP([FREQUENCIES no valid data])
 986 AT_DATA([empty.sps], [dnl
 987 data list notable list /x *.
 988 begin data.
 989 .
 990 .
 991 .
 992 end data.
 993
 994 FREQUENCIES
 995         /VARIABLES = x
 996         /STATISTICS = ALL
 997         .
 998 ])
 999
1000 AT_CHECK([pspp empty.sps -O format=csv], [0],  [dnl
1001 Table: Statistics
1002 ,,x
1003 N,Valid,0
1004 ,Missing,3
1005 Mean,,.  @&t@
1006 S.E. Mean,,.  @&t@
1007 Median,,.  @&t@
1008 Mode,,.  @&t@
1009 Std Dev,,.  @&t@
1010 Variance,,.  @&t@
1011 Kurtosis,,.  @&t@
1012 S.E. Kurt,,.  @&t@
1013 Skewness,,.  @&t@
1014 S.E. Skew,,.  @&t@
1015 Range,,.  @&t@
1016 Minimum,,.  @&t@
1017 Maximum,,.  @&t@
1018 Sum,,.  @&t@
1019
1020 Table: x
1021 ,,Frequency,Percent
1022 Missing,.  ,3,100.0%
1023 Total,,3,.0%
1024 ])
1025
1026 AT_CLEANUP
1027
1028
1029 AT_SETUP([FREQUENCIES histogram no valid cases])
1030 AT_DATA([empty.sps], [dnl
1031 data list notable list /x w *.
1032 begin data.
1033 1 .
1034 2 .
1035 3 .
1036 end data.
1037
1038 weight by w.
1039
1040 FREQUENCIES
1041         /VARIABLES = x
1042         /histogram
1043         .
1044 ])
1045
1046 AT_CHECK([pspp empty.sps -O format=csv], [0],  [ignore])
1047
1048 AT_CLEANUP
1049
1050 AT_SETUP([FREQUENCIES percentiles + histogram bug#48128])
1051 AT_DATA([bug.sps], [dnl
1052 SET FORMAT=F8.0.
1053
1054 INPUT PROGRAM.
1055         LOOP I=1 TO 10.
1056                 COMPUTE SCORE=EXP(NORMAL(1)).
1057                 END CASE.
1058         END LOOP.
1059         END FILE.
1060 END INPUT PROGRAM.
1061
1062 FREQUENCIES VARIABLES=SCORE
1063 /FORMAT=NOTABLE
1064 /STATISTICS=ALL
1065 /PERCENTILES=1 10 20 30 40 50 60 70 80 90 99
1066 /HISTOGRAM.
1067
1068 ])
1069
1070 AT_CHECK([pspp bug.sps], [0],  [ignore])
1071
1072 AT_CLEANUP
1073
1074
1075 AT_SETUP([FREQUENCIES vs. missing weights])
1076 AT_DATA([warn.sps], [dnl
1077 data list notable list /x w .
1078 begin data.
1079 1 1
1080 2 1
1081 1 1
1082 3 1
1083 3 .
1084 4 .
1085 end data.
1086
1087 weight by w.
1088
1089 frequencies /variables=x.
1090 ])
1091
1092 AT_CHECK([pspp warn.sps -O format=csv], [0],  [dnl
1093 "warn.sps:13: warning: FREQUENCIES: 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."
1094
1095 Table: Statistics
1096 ,,x
1097 N,Valid,4.00
1098 ,Missing,.00
1099 Mean,,1.75
1100 Std Dev,,.96
1101 Minimum,,1.00
1102 Maximum,,4.00
1103
1104 Table: x
1105 ,,Frequency,Percent,Valid Percent,Cumulative Percent
1106 Valid,1.00,2.00,50.0%,50.0%,50.0%
1107 ,2.00,1.00,25.0%,25.0%,75.0%
1108 ,3.00,1.00,25.0%,25.0%,100.0%
1109 ,4.00,.00,.0%,.0%,100.0%
1110 Total,,4.00,100.0%,,
1111 ])
1112
1113 AT_CLEANUP