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], [dnl
 134 Reading free-form data from INLINE.
 135 +--------+------+
 136 |Variable|Format|
 137 #========#======#
 138 |x       |F8.0  |
 139 |w       |F8.0  |
 140 +--------+------+
 141 ])
 142 AT_CLEANUP
 143
 144 # Tests for a bug which crashed PSPP when the median and a histogram
 145 # were both requested.
 146 AT_SETUP([FREQUENCIES median with histogram crash])
 147 AT_DATA([frequencies.sps], [dnl
 148 data list list notable /x.
 149 begin data.
 150 1
 151 end data.
 152
 153 frequencies /x /histogram /STATISTICS=median.
 154 ])
 155 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 156 Table: x
 157 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 158 ,1.00,1,100.00,100.00,100.00
 159 Total,,1,100.0,100.0,
 160
 161 Table: x
 162 N,Valid,1
 163 ,Missing,0
 164 S.E. Kurt,,.00
 165 ,50 (Median),1.00
 166 ])
 167 AT_CLEANUP
 168
 169 # Tests for a bug which caused FREQUENCIES following TEMPORARY to
 170 # crash (bug #11492).
 171 AT_SETUP([FREQUENCIES crash after TEMPORARY])
 172 AT_DATA([frequencies.sps],
 173   [DATA LIST LIST /SEX (A1) X *.
 174 BEGIN DATA.
 175 M 31
 176 F 21
 177 M 41
 178 F 31
 179 M 13
 180 F 12
 181 M 14
 182 F 13
 183 END DATA.
 184
 185
 186 TEMPORARY
 187 SELECT IF SEX EQ 'F'
 188 FREQUENCIES /X .
 189
 190 FINISH
 191 ])
 192 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 193   [Table: Reading free-form data from INLINE.
 194 Variable,Format
 195 SEX,A1
 196 X,F8.0
 197
 198 Table: X
 199 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 200 ,12.00,1,25.00,25.00,25.00
 201 ,13.00,1,25.00,25.00,50.00
 202 ,21.00,1,25.00,25.00,75.00
 203 ,31.00,1,25.00,25.00,100.00
 204 Total,,4,100.0,100.0,
 205
 206 Table: X
 207 N,Valid,4
 208 ,Missing,0
 209 Mean,,19.25
 210 Std Dev,,8.81
 211 Minimum,,12.00
 212 Maximum,,31.00
 213 ])
 214 AT_CLEANUP
 215
 216 m4_define([FREQUENCIES_NTILES_OUTPUT],
 217   [Table: x
 218 N,Valid,5
 219 ,Missing,0
 220 Mean,,3.00
 221 Std Dev,,1.58
 222 Minimum,,1.00
 223 Maximum,,5.00
 224 Percentiles,0,1.00
 225 ,25,2.00
 226 ,33,2.33
 227 ,50 (Median),3.00
 228 ,67,3.67
 229 ,75,4.00
 230 ,100,5.00
 231 ])
 232 AT_SETUP([FREQUENCIES basic percentiles])
 233 AT_DATA([frequencies.sps],
 234   [DATA LIST LIST notable /x * .
 235 BEGIN DATA.
 236 1
 237 2
 238 3
 239 4
 240 5
 241 END DATA.
 242
 243 FREQUENCIES
 244         VAR=x
 245         /FORMAT=NOTABLE
 246         /PERCENTILES = 0 25 33.333 50 66.666 75 100.
 247 ])
 248 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 249   [FREQUENCIES_NTILES_OUTPUT])
 250 AT_CLEANUP
 251
 252 AT_SETUP([FREQUENCIES basic n-tiles])
 253 AT_DATA([frequencies.sps],
 254   [DATA LIST LIST notable /x * .
 255 BEGIN DATA.
 256 1
 257 2
 258 3
 259 4
 260 5
 261 END DATA.
 262
 263 FREQUENCIES
 264         VAR=x
 265         /FORMAT=NOTABLE
 266         /NTILES = 3
 267         /NTILES = 4.
 268 ])
 269 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 270   [FREQUENCIES_NTILES_OUTPUT])
 271 AT_CLEANUP
 272
 273 AT_SETUP([FREQUENCIES compatibility percentiles])
 274 AT_DATA([frequencies.sps],
 275   [DATA LIST LIST notable /X * .
 276 BEGIN DATA.
 277 1
 278 2
 279 3
 280 4
 281 5
 282 END DATA.
 283
 284 FREQUENCIES
 285         VAR=x
 286         /ALGORITHM=COMPATIBLE
 287         /PERCENTILES = 0 25 50 75 100.
 288 ])
 289 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 290   [Table: X
 291 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 292 ,1.00,1,20.00,20.00,20.00
 293 ,2.00,1,20.00,20.00,40.00
 294 ,3.00,1,20.00,20.00,60.00
 295 ,4.00,1,20.00,20.00,80.00
 296 ,5.00,1,20.00,20.00,100.00
 297 Total,,5,100.0,100.0,
 298
 299 Table: X
 300 N,Valid,5
 301 ,Missing,0
 302 Mean,,3.00
 303 Std Dev,,1.58
 304 Minimum,,1.00
 305 Maximum,,5.00
 306 Percentiles,0,1.00
 307 ,25,1.50
 308 ,50 (Median),3.00
 309 ,75,4.50
 310 ,100,5.00
 311 ])
 312 AT_CLEANUP
 313
 314 AT_SETUP([FREQUENCIES enhanced percentiles])
 315 AT_DATA([frequencies.sps],
 316   [DATA LIST LIST notable /X * .
 317 BEGIN DATA.
 318 1
 319 2
 320 3
 321 4
 322 5
 323 END DATA.
 324
 325 FREQUENCIES
 326         VAR=x
 327         /PERCENTILES = 0 25 50 75 100.
 328 ])
 329 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 330   [Table: X
 331 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 332 ,1.00,1,20.00,20.00,20.00
 333 ,2.00,1,20.00,20.00,40.00
 334 ,3.00,1,20.00,20.00,60.00
 335 ,4.00,1,20.00,20.00,80.00
 336 ,5.00,1,20.00,20.00,100.00
 337 Total,,5,100.0,100.0,
 338
 339 Table: X
 340 N,Valid,5
 341 ,Missing,0
 342 Mean,,3.00
 343 Std Dev,,1.58
 344 Minimum,,1.00
 345 Maximum,,5.00
 346 Percentiles,0,1.00
 347 ,25,2.00
 348 ,50 (Median),3.00
 349 ,75,4.00
 350 ,100,5.00
 351 ])
 352 AT_CLEANUP
 353
 354 AT_SETUP([FREQUENCIES enhanced percentiles, weighted])
 355 AT_DATA([frequencies.sps],
 356   [DATA LIST LIST notable /X * F *.
 357 BEGIN DATA.
 358 1 2
 359 2 2
 360 3 2
 361 4 1
 362 4 1
 363 5 1
 364 5 1
 365 END DATA.
 366
 367 WEIGHT BY f.
 368
 369 FREQUENCIES
 370         VAR=x
 371         /PERCENTILES = 0 25 50 75 100.
 372 ])
 373 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 374   [Table: X
 375 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 376 ,1.00,2.00,20.00,20.00,20.00
 377 ,2.00,2.00,20.00,20.00,40.00
 378 ,3.00,2.00,20.00,20.00,60.00
 379 ,4.00,2.00,20.00,20.00,80.00
 380 ,5.00,2.00,20.00,20.00,100.00
 381 Total,,10.00,100.0,100.0,
 382
 383 Table: X
 384 N,Valid,10.00
 385 ,Missing,.00
 386 Mean,,3.00
 387 Std Dev,,1.49
 388 Minimum,,1.00
 389 Maximum,,5.00
 390 Percentiles,0,1.00
 391 ,25,2.00
 392 ,50 (Median),3.00
 393 ,75,4.00
 394 ,100,5.00
 395 ])
 396 AT_CLEANUP
 397
 398 AT_SETUP([FREQUENCIES enhanced percentiles, weighted (2)])
 399 AT_DATA([frequencies.sps],
 400   [DATA LIST LIST notable /X * F *.
 401 BEGIN DATA.
 402 1 1
 403 3 2
 404 4 1
 405 5 1
 406 5 1
 407 END DATA.
 408
 409 WEIGHT BY f.
 410
 411 FREQUENCIES
 412         VAR=x
 413         /PERCENTILES = 0 25 50 75 100.
 414 ])
 415 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 416   [Table: X
 417 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 418 ,1.00,1.00,16.67,16.67,16.67
 419 ,3.00,2.00,33.33,33.33,50.00
 420 ,4.00,1.00,16.67,16.67,66.67
 421 ,5.00,2.00,33.33,33.33,100.00
 422 Total,,6.00,100.0,100.0,
 423
 424 Table: X
 425 N,Valid,6.00
 426 ,Missing,.00
 427 Mean,,3.50
 428 Std Dev,,1.52
 429 Minimum,,1.00
 430 Maximum,,5.00
 431 Percentiles,0,1.00
 432 ,25,3.00
 433 ,50 (Median),3.50
 434 ,75,4.75
 435 ,100,5.00
 436 ])
 437 AT_CLEANUP
 438
 439 dnl Data for this test case from Fabio Bordignon <bordignon@demos.it>.
 440 AT_SETUP([FREQUENCIES enhanced percentiles, weighted (3)])
 441 AT_DATA([frequencies.sps],
 442   [DATA LIST LIST notable /X * F *.
 443 BEGIN DATA.
 444 1 7
 445 2 16
 446 3 12
 447 4 5
 448 END DATA.
 449
 450 WEIGHT BY f.
 451
 452 FREQUENCIES
 453         VAR=x
 454         /PERCENTILES = 0 25 50 75 100.
 455 ])
 456 AT_CHECK([pspp -O format=csv frequencies.sps], [0], [dnl
 457 Table: X
 458 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 459 ,1.00,7.00,17.50,17.50,17.50
 460 ,2.00,16.00,40.00,40.00,57.50
 461 ,3.00,12.00,30.00,30.00,87.50
 462 ,4.00,5.00,12.50,12.50,100.00
 463 Total,,40.00,100.0,100.0,
 464
 465 Table: X
 466 N,Valid,40.00
 467 ,Missing,.00
 468 Mean,,2.38
 469 Std Dev,,.93
 470 Minimum,,1.00
 471 Maximum,,4.00
 472 Percentiles,0,1.00
 473 ,25,2.00
 474 ,50 (Median),2.00
 475 ,75,3.00
 476 ,100,4.00
 477 ])
 478 AT_CLEANUP
 479
 480 AT_SETUP([FREQUENCIES enhanced percentiles, weighted, missing values])
 481 AT_DATA([frequencies.sps],
 482   [DATA LIST LIST notable /X * F *.
 483 BEGIN DATA.
 484 1 1
 485 3 2
 486 4 1
 487 5 1
 488 5 1
 489 99 4
 490 END DATA.
 491
 492 MISSING VALUE x (99.0) .
 493 WEIGHT BY f.
 494
 495 FREQUENCIES
 496         VAR=x
 497         /PERCENTILES = 0 25 50 75 100.
 498 ])
 499 AT_CHECK([pspp -O format=csv frequencies.sps], [0],
 500   [Table: X
 501 Value Label,Value,Frequency,Percent,Valid Percent,Cum Percent
 502 ,1.00,1.00,10.00,16.67,16.67
 503 ,3.00,2.00,20.00,33.33,50.00
 504 ,4.00,1.00,10.00,16.67,66.67
 505 ,5.00,2.00,20.00,33.33,100.00
 506 ,99.00,4.00,40.00,Missing,
 507 Total,,10.00,100.0,100.0,
 508
 509 Table: X
 510 N,Valid,6.00
 511 ,Missing,4.00
 512 Mean,,3.50
 513 Std Dev,,1.52
 514 Minimum,,1.00
 515 Maximum,,5.00
 516 Percentiles,0,1.00
 517 ,25,3.00
 518 ,50 (Median),3.50
 519 ,75,4.75
 520 ,100,5.00
 521 ])
 522 AT_CLEANUP