statist-1.4.2.tar.gz: .../run_comparison.sh

"Fossies" - the Fresh Open Source Software Archive

Member "statist-1.4.2/tools/run_comparison.sh" (10 Sep 2006, 5182 Bytes) of package /linux/privat/old/statist-1.4.2.tar.gz:

As a special service "Fossies" has tried to format the requested source page into HTML format using (guessed) Bash source code syntax highlighting (style: standard) with prefixed line numbers and code folding option. Alternatively you can here view or download the uninterpreted source code file.

1 #!/bin/bash 2 3 # This file is part of statist 4 # 5 # It is distributed under the GNU General Public License. 6 # See the file COPYING for details. 7 # 8 # (C) 2006 Jakson Alves de Aquino <jalvesaq@gmail.com> 9 # 10 # $Id: run_comparison.sh,v 1.3 2006/09/10 02:42:12 jakson Exp $ 11 12 # Warning: This script is the worst possible example of how to do statistics. 13 # The analyzes carried here have the only goal of comparing two different 14 # versions of statist. 15 16 # This script will run two different versions of statist to check whether the 17 # new version still produces the correct results. The old version is the one 18 # tagged as "bernhard-last-maint-2005" (here called $statist_102) and the new 19 # one is the 1.4.1. You have to compile these versions of statist and write 20 # the path to them in the variables below. The path to statist "examples" 21 # subdirectory is necessary because the file city.csv will be used. 22 23 statist_102="./statist-1.0.2" 24 statist_141="../src/statist" 25 examples_dir="../examples" 26 27 LANG=C 28 LANGUAGE=C 29 LC_ALL=C 30 31 # Before running statist-1.0.2 we must create four new data files to run 32 # analysis that require the same number of rows. Statist-1.0.2 has the option 33 # -delrow, but, alone, it isn't enough. 34 35 # NOTE 1: We can't use statist-1.0.2 to "export a file" with the columns 36 # because the rows with missing values will be dislocated. Example: 37 38 # 34 44 34 44 39 # 35 M will become 35 53 40 # 36 53 36 . 41 42 # As you can see, the "53" is no longer aligned with "36". 43 44 # NOTE 2: statist-1.0.2 would be unable to read the two columns that it had 45 # just exported due to the "dot" at the third column. The correct (for statist) 46 # would be to put a "M" in the second line, not a dot at the third one. So, 47 # let's use awk: 48 49 sed -e 's/age/#%age/' $examples_dir/city.csv > city.dat 50 awk '{if($2 == "sex") {print "#%sex god"} else if($2 == 1 || $2 == 0) print $2, $6}' city.dat > zero_one.dat 51 awk '{print $1, $7}' city.dat > two_columns.dat 52 awk '{print $1, $7, $8}' city.dat > three_columns.dat 53 awk '{print $1, $2, $4, $7, $8}' city.dat > five_columns.dat 54 55 56 # Now we'll run statist-1.0.2 six times. 57 58 # FIRST: We'll use two_columns.dat to run all menu items that require two 59 # columns with exactly the same number of rows. Note: In a real research we 60 # wouldn't have this privilege. We should use awk to save all combinations of 61 # two columns that we were interested in. 62 63 # SECOND: We'll use three_columns.dat to run all menu items that require 64 # three columns with exactly the same number of rows. 65 66 # THIRD: We'll run partial linear correlation with 5 items. 67 68 # FORTH: We'll run Chi^2-fourfold-test (zero_one.dat). 69 70 # FIFTH: We'll run all remaining menu items. 71 72 # SIXTH: We'll run the probit analysis. 73 74 # menu choices for the first run: 75 echo '2 76 1 77 age 78 deg 79 2 80 age 81 deg 82 5 83 age 84 deg 85 5 86 0 87 3 88 2 89 age 90 deg 91 8 92 age 93 deg 94 0 95 0 96 ' > run_it 97 98 $statist_102 -delrow -silent -noplot two_columns.dat < run_it > st102a 99 100 101 # menu choices for the second run (three_columns): 102 echo '2 103 3 104 3 105 inc 106 deg 107 age 108 9 109 3 110 inc 111 deg 112 age 113 10 114 3 115 inc 116 deg 117 age 118 5 119 0 120 3 121 5 122 3 123 inc 124 deg 125 age 126 0 127 0 128 ' > run_it 129 130 $statist_102 -delrow -silent -noplot three_columns.dat < run_it > st102b 131 132 133 # menu choices for the third run (five_columns): 134 echo '2 135 4 136 5 137 inc 138 age 139 deg 140 hap 141 sex 142 6 143 5 144 inc 145 age 146 deg 147 hap 148 sex 149 7 150 5 151 inc 152 age 153 deg 154 hap 155 sex 156 0 157 0 158 ' > run_it 159 160 $statist_102 -silent -delrow -noplot five_columns.dat < run_it > st102c 161 162 163 # menu choices for the fourth run (Chi^2): 164 echo '3 165 4 166 sex 167 god 168 0 169 0 170 ' > run_it 171 172 $statist_102 -silent -delrow -noplot zero_one.dat < run_it > st102d 173 174 175 # menu choices for the fifth run (city.dat): 176 echo '2 177 8 178 sex 179 age 180 0 181 3 182 1 183 deg 184 inc 185 6 186 deg 187 inc 188 7 189 3 190 deg 191 inc 192 age 193 9 194 deg 195 10 196 sex 197 god 198 0 199 4 200 1 201 age 202 203 204 205 3 206 inc 207 n 208 4 209 age 210 0 211 0' > run_it 212 213 $statist_102 -silent -noplot city.dat < run_it > st102e 214 215 # menu choices for the sixth run (probit): 216 echo '4 217 2 218 y 219 dose 220 ef 221 2 222 N 223 dos 224 num 225 ef 226 0 227 0 228 ' > run_it 229 230 $statist_102 -silent -noplot $examples_dir/probit.dat < run_it > st102f 231 232 cat st102* > result_statist-1.0.2 233 rm st102* 234 235 236 ########################################################################## 237 # Now, finally, we'll write the menu choices for statist-1.4.1 and run it. 238 # Note that statist-1.4.1 doesn't have (and doesn't need) the option -delrow. 239 240 echo '2 241 1 242 age 243 deg 244 2 245 age 246 deg 247 5 248 age 249 deg 250 5 251 0 252 3 253 2 254 age 255 deg 256 8 257 age 258 deg 259 0 260 2 261 3 262 inc 263 deg 264 age 265 266 9 267 inc 268 deg 269 age 270 271 10 272 inc 273 deg 274 age 275 276 5 277 0 278 3 279 5 280 inc 281 deg 282 age 283 284 y 285 0 286 2 287 4 288 inc 289 age 290 deg 291 hap 292 sex 293 6 294 inc 295 age 296 deg 297 hap 298 sex 299 300 7 301 inc 302 age 303 deg 304 hap 305 sex 306 307 0 308 3 309 4 310 sex 311 god 312 0 313 2 314 8 315 sex 316 age 317 0 318 3 319 1 320 deg 321 inc 322 6 323 deg 324 inc 325 7 326 deg 327 inc 328 age 329 330 9 331 deg 332 10 333 sex 334 god 335 0 336 4 337 1 338 age 339 340 341 342 3 343 inc 344 n 345 4 346 age 347 0 348 0' > run_it 349 350 $statist_141 --na-string "M" --noplot --silent city.dat < run_it > st140a 351 352 353 # menu choices for probit: 354 echo '4 355 2 356 y 357 dose 358 ef 359 2 360 361 dos 362 num 363 ef 364 0 365 0 366 ' > run_it 367 368 369 $statist_141 --noplot --silent $examples_dir/probit.dat < run_it > st140b 370 371 cat st140* > result_statist-1.4.1 372 373 rm st140* city.dat run_it five_columns.dat three_columns.dat two_columns.dat zero_one.dat 374 375 # Note: It's expected different results for "Partial linear correlation" with 5 376 # variables because statist-1.0.2 was using an unintialized variable. This bug 377 # was fixed in statist-1.3.1 (thanks to valgrind). 378 379 echo 380 echo "The results were saved in the files" 381 echo "result_statist-1.0.2 and result_statist-1.4.1." 382 echo 383 exit 0 384 385