tests/language/stats/frequencies.at

   1 AT_BANNER([FREQUENCIES procedure])
   2
   3 AT_SETUP([FREQUENCIES string variable crash])
   4 AT_DATA([frequencies.sps],
   5   [DATA LIST FREE/
   6    name  (A8) value * quantity .
   7 BEGIN DATA.
   8 Cables 829 3
   9 END DATA.
  10 EXECUTE.
  11
  12 FREQUENCIES /VAR = name.
  13 ])
  14 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
  15   [Table: name
  16 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
  17 ,Cables  ,1,100.00,100.00,100.00
  18 Total,,1,100.0,100.0,
  19 ])
  20 AT_CLEANUP
  21
  22 # Tests for a bug where pspp would crash if two FREQUENCIES commands
  23 # existed in a input file.
  24 AT_SETUP([FREQUENCIES two runs crash])
  25 AT_DATA([frequencies.sps],
  26   [data list free /v1 v2.
  27 begin data.
  28 0 1
  29 2 3
  30 4 5
  31 3 4
  32 end data.
  33
  34 frequencies v1 v2/statistics=none.
  35 frequencies v1 v2/statistics=none.
  36 ])
  37 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
  38   [Table: v1
  39 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
  40 ,.00,1,25.00,25.00,25.00
  41 ,2.00,1,25.00,25.00,50.00
  42 ,3.00,1,25.00,25.00,75.00
  43 ,4.00,1,25.00,25.00,100.00
  44 Total,,4,100.0,100.0,
  45
  46 Table: v2
  47 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
  48 ,1.00,1,25.00,25.00,25.00
  49 ,3.00,1,25.00,25.00,50.00
  50 ,4.00,1,25.00,25.00,75.00
  51 ,5.00,1,25.00,25.00,100.00
  52 Total,,4,100.0,100.0,
  53
  54 Table: v1
  55 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
  56 ,.00,1,25.00,25.00,25.00
  57 ,2.00,1,25.00,25.00,50.00
  58 ,3.00,1,25.00,25.00,75.00
  59 ,4.00,1,25.00,25.00,100.00
  60 Total,,4,100.0,100.0,
  61
  62 Table: v2
  63 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
  64 ,1.00,1,25.00,25.00,25.00
  65 ,3.00,1,25.00,25.00,50.00
  66 ,4.00,1,25.00,25.00,75.00
  67 ,5.00,1,25.00,25.00,100.00
  68 Total,,4,100.0,100.0,
  69 ])
  70 AT_CLEANUP
  71
  72 # Tests for a bug where PSPP would crash when a FREQUENCIES command
  73 # was used with the HTML output driver.
  74 AT_SETUP([FREQUENCIES HTML output crash])
  75 AT_DATA([frequencies.sps],
  76   [data list free /v1 v2.
  77 begin data.
  78 0 1
  79 2 3
  80 4 5
  81 3 4
  82 end data.
  83
  84 list.
  85
  86 frequencies v1/statistics=none.
  87 ])
  88 AT_CHECK([pspp -o - -O format=csv -o pspp.html frequencies.sps], [0],
  89   [Table: Data List
  90 v1,v2
  91 .00,1.00
  92 2.00,3.00
  93 4.00,5.00
  94 3.00,4.00
  95
  96 Table: v1
  97 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
  98 ,.00,1,25.00,25.00,25.00
  99 ,2.00,1,25.00,25.00,50.00
 100 ,3.00,1,25.00,25.00,75.00
 101 ,4.00,1,25.00,25.00,100.00
 102 Total,,4,100.0,100.0,
 103 ])
 104 AT_CHECK([test -s pspp.html])
 105 AT_CLEANUP
 106
 107 # Tests for a bug which crashed PSPP when a piechart with too many
 108 # segments was requested.
 109 AT_SETUP([FREQUENCIES pie chart crash])
 110 AT_DATA([frequencies.sps],
 111   [data list list /x * w *.
 112 begin data.
 113 1  4
 114 34 10
 115 -9 15
 116 232 6
 117 11  4
 118 134 1
 119 9  5
 120 32 16
 121 -2 6
 122 2  16
 123 20  6
 124 end data.
 125
 126 weight by w.
 127
 128 frequencies /x /format=notable /statistics=none
 129         /piechart.
 130 ])
 131 # Cannot use the CSV driver for this because it does not output charts
 132 # at all.
 133 AT_CHECK([pspp frequencies.sps], [0],
 134   [DATA LIST
 135
 136 Reading free-form data from INLINE.
 137 +--------+------+
 138 |Variable|Format|
 139 #========#======#
 140 |x       |F8.0  |
 141 |w       |F8.0  |
 142 +--------+------+
 143
 144 BEGIN DATA
 145
 146 WEIGHT
 147
 148 FREQUENCIES
 149 ])
 150 AT_CLEANUP
 151
 152 # Tests for a bug which crashed PSPP when the median and a histogram
 153 # were both requested.
 154 AT_SETUP([FREQUENCIES median with histogram crash])
 155 AT_DATA([frequencies.sps], [dnl
 156 data list list notable /x.
 157 begin data.
 158 1
 159 end data.
 160
 161 frequencies /x /histogram /STATISTICS=median.
 162 ])
 163 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 164 Table: x
 165 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 166 ,1.00,1,100.00,100.00,100.00
 167 Total,,1,100.0,100.0,
 168
 169 Table: x
 170 N,Valid,1
 171 ,Missing,0
 172 S.E. Kurt,,.00
 173 ,50 (Median),1.00
 174 ])
 175 AT_CLEANUP
 176
 177 # Tests for a bug which caused FREQUENCIES following TEMPORARY to
 178 # crash (bug #11492).
 179 AT_SETUP([FREQUENCIES crash after TEMPORARY])
 180 AT_DATA([frequencies.sps],
 181   [DATA LIST LIST /SEX (A1) X *.
 182 BEGIN DATA.
 183 M 31
 184 F 21
 185 M 41
 186 F 31
 187 M 13
 188 F 12
 189 M 14
 190 F 13
 191 END DATA.
 192
 193
 194 TEMPORARY
 195 SELECT IF SEX EQ 'F'
 196 FREQUENCIES /X .
 197
 198 FINISH
 199 ])
 200 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 201   [Table: Reading free-form data from INLINE.
 202 Variable,Format
 203 SEX,A1
 204 X,F8.0
 205
 206 Table: X
 207 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 208 ,12.00,1,25.00,25.00,25.00
 209 ,13.00,1,25.00,25.00,50.00
 210 ,21.00,1,25.00,25.00,75.00
 211 ,31.00,1,25.00,25.00,100.00
 212 Total,,4,100.0,100.0,
 213
 214 Table: X
 215 N,Valid,4
 216 ,Missing,0
 217 Mean,,19.25
 218 Std Dev,,8.81
 219 Minimum,,12.00
 220 Maximum,,31.00
 221 ])
 222 AT_CLEANUP
 223
 224 m4_define([FREQUENCIES_NTILES_OUTPUT],
 225   [Table: x
 226 N,Valid,5
 227 ,Missing,0
 228 Mean,,3.00
 229 Std Dev,,1.58
 230 Minimum,,1.00
 231 Maximum,,5.00
 232 Percentiles,0,1.00
 233 ,25,2.00
 234 ,33,2.33
 235 ,50 (Median),3.00
 236 ,67,3.67
 237 ,75,4.00
 238 ,100,5.00
 239 ])
 240 AT_SETUP([FREQUENCIES basic percentiles])
 241 AT_DATA([frequencies.sps],
 242   [DATA LIST LIST notable /x * .
 243 BEGIN DATA.
 244 1
 245 2
 246 3
 247 4
 248 5
 249 END DATA.
 250
 251 FREQUENCIES
 252         VAR=x
 253         /FORMAT=NOTABLE
 254         /PERCENTILES = 0 25 33.333 50 66.666 75 100.
 255 ])
 256 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 257   [FREQUENCIES_NTILES_OUTPUT])
 258 AT_CLEANUP
 259
 260 AT_SETUP([FREQUENCIES basic n-tiles])
 261 AT_DATA([frequencies.sps],
 262   [DATA LIST LIST notable /x * .
 263 BEGIN DATA.
 264 1
 265 2
 266 3
 267 4
 268 5
 269 END DATA.
 270
 271 FREQUENCIES
 272         VAR=x
 273         /FORMAT=NOTABLE
 274         /NTILES = 3
 275         /NTILES = 4.
 276 ])
 277 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 278   [FREQUENCIES_NTILES_OUTPUT])
 279 AT_CLEANUP
 280
 281 AT_SETUP([FREQUENCIES compatibility percentiles])
 282 AT_DATA([frequencies.sps],
 283   [DATA LIST LIST notable /X * .
 284 BEGIN DATA.
 285 1
 286 2
 287 3
 288 4
 289 5
 290 END DATA.
 291
 292 FREQUENCIES
 293         VAR=x
 294         /ALGORITHM=COMPATIBLE
 295         /PERCENTILES = 0 25 50 75 100.
 296 ])
 297 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 298   [Table: X
 299 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 300 ,1.00,1,20.00,20.00,20.00
 301 ,2.00,1,20.00,20.00,40.00
 302 ,3.00,1,20.00,20.00,60.00
 303 ,4.00,1,20.00,20.00,80.00
 304 ,5.00,1,20.00,20.00,100.00
 305 Total,,5,100.0,100.0,
 306
 307 Table: X
 308 N,Valid,5
 309 ,Missing,0
 310 Mean,,3.00
 311 Std Dev,,1.58
 312 Minimum,,1.00
 313 Maximum,,5.00
 314 Percentiles,0,1.00
 315 ,25,1.50
 316 ,50 (Median),3.00
 317 ,75,4.50
 318 ,100,5.00
 319 ])
 320 AT_CLEANUP
 321
 322 AT_SETUP([FREQUENCIES enhanced percentiles])
 323 AT_DATA([frequencies.sps],
 324   [DATA LIST LIST notable /X * .
 325 BEGIN DATA.
 326 1
 327 2
 328 3
 329 4
 330 5
 331 END DATA.
 332
 333 FREQUENCIES
 334         VAR=x
 335         /PERCENTILES = 0 25 50 75 100.
 336 ])
 337 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 338   [Table: X
 339 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 340 ,1.00,1,20.00,20.00,20.00
 341 ,2.00,1,20.00,20.00,40.00
 342 ,3.00,1,20.00,20.00,60.00
 343 ,4.00,1,20.00,20.00,80.00
 344 ,5.00,1,20.00,20.00,100.00
 345 Total,,5,100.0,100.0,
 346
 347 Table: X
 348 N,Valid,5
 349 ,Missing,0
 350 Mean,,3.00
 351 Std Dev,,1.58
 352 Minimum,,1.00
 353 Maximum,,5.00
 354 Percentiles,0,1.00
 355 ,25,2.00
 356 ,50 (Median),3.00
 357 ,75,4.00
 358 ,100,5.00
 359 ])
 360 AT_CLEANUP
 361
 362 AT_SETUP([FREQUENCIES enhanced percentiles, weighted])
 363 AT_DATA([frequencies.sps],
 364   [DATA LIST LIST notable /X * F *.
 365 BEGIN DATA.
 366 1 2
 367 2 2
 368 3 2
 369 4 1
 370 4 1
 371 5 1
 372 5 1
 373 END DATA.
 374
 375 WEIGHT BY f.
 376
 377 FREQUENCIES
 378         VAR=x
 379         /PERCENTILES = 0 25 50 75 100.
 380 ])
 381 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 382   [Table: X
 383 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 384 ,1.00,2.00,20.00,20.00,20.00
 385 ,2.00,2.00,20.00,20.00,40.00
 386 ,3.00,2.00,20.00,20.00,60.00
 387 ,4.00,2.00,20.00,20.00,80.00
 388 ,5.00,2.00,20.00,20.00,100.00
 389 Total,,10.00,100.0,100.0,
 390
 391 Table: X
 392 N,Valid,10.00
 393 ,Missing,.00
 394 Mean,,3.00
 395 Std Dev,,1.49
 396 Minimum,,1.00
 397 Maximum,,5.00
 398 Percentiles,0,1.00
 399 ,25,2.00
 400 ,50 (Median),3.00
 401 ,75,4.00
 402 ,100,5.00
 403 ])
 404 AT_CLEANUP
 405
 406 AT_SETUP([FREQUENCIES enhanced percentiles, weighted (2)])
 407 AT_DATA([frequencies.sps],
 408   [DATA LIST LIST notable /X * F *.
 409 BEGIN DATA.
 410 1 1
 411 3 2
 412 4 1
 413 5 1
 414 5 1
 415 END DATA.
 416
 417 WEIGHT BY f.
 418
 419 FREQUENCIES
 420         VAR=x
 421         /PERCENTILES = 0 25 50 75 100.
 422 ])
 423 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 424   [Table: X
 425 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 426 ,1.00,1.00,16.67,16.67,16.67
 427 ,3.00,2.00,33.33,33.33,50.00
 428 ,4.00,1.00,16.67,16.67,66.67
 429 ,5.00,2.00,33.33,33.33,100.00
 430 Total,,6.00,100.0,100.0,
 431
 432 Table: X
 433 N,Valid,6.00
 434 ,Missing,.00
 435 Mean,,3.50
 436 Std Dev,,1.52
 437 Minimum,,1.00
 438 Maximum,,5.00
 439 Percentiles,0,1.00
 440 ,25,3.00
 441 ,50 (Median),3.50
 442 ,75,4.75
 443 ,100,5.00
 444 ])
 445 AT_CLEANUP
 446
 447 dnl Data for this test case from Fabio Bordignon <bordignon@demos.it>.
 448 AT_SETUP([FREQUENCIES enhanced percentiles, weighted (3)])
 449 AT_DATA([frequencies.sps],
 450   [DATA LIST LIST notable /X * F *.
 451 BEGIN DATA.
 452 1 7
 453 2 16
 454 3 12
 455 4 5
 456 END DATA.
 457
 458 WEIGHT BY f.
 459
 460 FREQUENCIES
 461         VAR=x
 462         /PERCENTILES = 0 25 50 75 100.
 463 ])
 464 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 465 Table: X
 466 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 467 ,1.00,7.00,17.50,17.50,17.50
 468 ,2.00,16.00,40.00,40.00,57.50
 469 ,3.00,12.00,30.00,30.00,87.50
 470 ,4.00,5.00,12.50,12.50,100.00
 471 Total,,40.00,100.0,100.0,
 472
 473 Table: X
 474 N,Valid,40.00
 475 ,Missing,.00
 476 Mean,,2.38
 477 Std Dev,,.93
 478 Minimum,,1.00
 479 Maximum,,4.00
 480 Percentiles,0,1.00
 481 ,25,2.00
 482 ,50 (Median),2.00
 483 ,75,3.00
 484 ,100,4.00
 485 ])
 486 AT_CLEANUP
 487
 488 AT_SETUP([FREQUENCIES enhanced percentiles, weighted, missing values])
 489 AT_DATA([frequencies.sps],
 490   [DATA LIST LIST notable /X * F *.
 491 BEGIN DATA.
 492 1 1
 493 3 2
 494 4 1
 495 5 1
 496 5 1
 497 99 4
 498 END DATA.
 499
 500 MISSING VALUE x (99.0) .
 501 WEIGHT BY f.
 502
 503 FREQUENCIES
 504         VAR=x
 505         /PERCENTILES = 0 25 50 75 100.
 506 ])
 507 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 508   [Table: X
 509 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 510 ,1.00,1.00,10.00,16.67,16.67
 511 ,3.00,2.00,20.00,33.33,50.00
 512 ,4.00,1.00,10.00,16.67,66.67
 513 ,5.00,2.00,20.00,33.33,100.00
 514 ,99.00,4.00,40.00,Missing,
 515 Total,,10.00,100.0,100.0,
 516
 517 Table: X
 518 N,Valid,6.00
 519 ,Missing,4.00
 520 Mean,,3.50
 521 Std Dev,,1.52
 522 Minimum,,1.00
 523 Maximum,,5.00
 524 Percentiles,0,1.00
 525 ,25,3.00
 526 ,50 (Median),3.50
 527 ,75,4.75
 528 ,100,5.00
 529 ])
 530 AT_CLEANUP