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