/usr/share/pyshared/mvpa/datasets/base.py is in python-mvpa 0.4.8-3.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 | # emacs: -*- mode: python; py-indent-offset: 4; indent-tabs-mode: nil -*-
# vi: set ft=python sts=4 ts=4 sw=4 et:
### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ##
#
# See COPYING file distributed along with the PyMVPA package for the
# copyright and license terms.
#
### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ##
"""Dataset container"""
__docformat__ = 'restructuredtext'
import operator
import random
import mvpa.support.copy as copy
import numpy as N
# Sooner or later Dataset would become ClassWithCollections as well, but for
# now just an object -- thus commenting out tentative changes
#
#XXX from mvpa.misc.state import ClassWithCollections, SampleAttribute
from mvpa.misc.exceptions import DatasetError
from mvpa.misc.support import idhash as idhash_
from mvpa.base.dochelpers import enhancedDocString, table2string
from mvpa.base import warning
if __debug__:
from mvpa.base import debug
def _validate_indexes_uniq_sorted(seq, fname, item):
"""Helper function to validate that seq contains unique sorted values
"""
if operator.isSequenceType(seq):
seq_unique = N.unique(seq)
if len(seq) != len(seq_unique):
warning("%s() operates only with indexes for %s without"
" repetitions. Repetitions were removed."
% (fname, item))
if N.any(N.sort(seq) != seq_unique):
warning("%s() does not guarantee the original order"
" of selected %ss. Use selectSamples() and "
" selectFeatures(sort=False) instead" % (fname, item))
#XXX class Dataset(ClassWithCollections):
class Dataset(object):
"""*The* Dataset.
This class provides a container to store all necessary data to
perform MVPA analyses. These are the data samples, as well as the
labels associated with the samples. Additionally, samples can be
grouped into chunks.
:Groups:
- `Creators`: `__init__`, `selectFeatures`, `selectSamples`,
`applyMapper`
- `Mutators`: `permuteLabels`
Important: labels assumed to be immutable, i.e. no one should modify
them externally by accessing indexed items, ie something like
``dataset.labels[1] += 100`` should not be used. If a label has
to be modified, full copy of labels should be obtained, operated on,
and assigned back to the dataset, otherwise dataset.uniquelabels
would not work. The same applies to any other attribute which has
corresponding unique* access property.
"""
# XXX Notes about migration to use Collections to store data and
# attributes for samples, features, and dataset itself:
# changes:
# _data -> s_attr collection (samples attributes)
# _dsattr -> ds_attr collection
# f_attr collection (features attributes)
# static definition to track which unique attributes
# have to be reset/recomputed whenever anything relevant
# changes
# unique{labels,chunks} become a part of dsattr
_uniqueattributes = []
"""Unique attributes associated with the data"""
_registeredattributes = []
"""Registered attributes (stored in _data)"""
_requiredattributes = ['samples', 'labels']
"""Attributes which have to be provided to __init__, or otherwise
no default values would be assumed and construction of the
instance would fail"""
#XXX _ATTRIBUTE_COLLECTIONS = [ 's_attr', 'f_attr', 'ds_attr' ]
#XXX """Assure those 3 collections to be present in all datasets"""
#XXX
#XXX samples__ = SampleAttribute(doc="Samples data. 0th index is time", hasunique=False) # XXX
#XXX labels__ = SampleAttribute(doc="Labels for the samples", hasunique=True)
#XXX chunks__ = SampleAttribute(doc="Chunk identities for the samples", hasunique=True)
#XXX # samples ids (already unique by definition)
#XXX origids__ = SampleAttribute(doc="Chunk identities for the samples", hasunique=False)
def __init__(self,
# for copy constructor
data=None,
dsattr=None,
# automatic dtype conversion
dtype=None,
# new instances
samples=None,
labels=None,
labels_map=None,
chunks=None,
origids=None,
# flags
check_data=True,
copy_samples=False,
copy_data=True,
copy_dsattr=True):
"""Initialize dataset instance
There are basically two different way to create a dataset:
1. Create a new dataset from samples and sample attributes. In
this mode a two-dimensional `ndarray` has to be passed to the
`samples` keyword argument and the corresponding samples
attributes are provided via the `labels` and `chunks`
arguments.
2. Copy contructor mode
The second way is used internally to perform quick coyping
of datasets, e.g. when performing feature selection. In this
mode and the two dictionaries (`data` and `dsattr`) are
required. For performance reasons this mode bypasses most of
the sanity check performed by the previous mode, as for
internal operations data integrity is assumed.
:Parameters:
data : dict
Dictionary with an arbitrary number of entries. The value for
each key in the dict has to be an ndarray with the
same length as the number of rows in the samples array.
A special entry in this dictionary is 'samples', a 2d array
(samples x features). A shallow copy is stored in the object.
dsattr : dict
Dictionary of dataset attributes. An arbitrary number of
arbitrarily named and typed objects can be stored here. A
shallow copy of the dictionary is stored in the object.
dtype: type | None
If None -- do not change data type if samples
is an ndarray. Otherwise convert samples to dtype.
:Keywords:
samples : ndarray
2d array (samples x features)
labels
An array or scalar value defining labels for each samples.
Generally `labels` should be numeric, unless `labels_map`
is used
labels_map : None or bool or dict
Map original labels into numeric labels. If True, the
mapping is computed if labels are literal. If is False,
no mapping is computed. If dict instance -- provided
mapping is verified and applied. If you want to have
labels_map just be present given already numeric labels,
just assign labels_map dictionary to existing dataset
instance
chunks
An array or scalar value defining chunks for each sample
Each of the Keywords arguments overwrites what is/might be
already in the `data` container.
"""
#XXX ClassWithCollections.__init__(self)
# see if data and dsattr are none, if so, make them empty dicts
if data is None:
data = {}
if dsattr is None:
dsattr = {}
# initialize containers; default values are empty dicts
# always make a shallow copy of what comes in, otherwise total chaos
# is likely to happen soon
if copy_data:
# deep copy (cannot use copy.deepcopy, because samples is an
# exception
# but shallow copy first to get a shared version of the data in
# any case
lcl_data = data.copy()
for k, v in data.iteritems():
# skip copying samples if requested
if k == 'samples' and not copy_samples:
continue
lcl_data[k] = v.copy()
else:
# shallow copy
# XXX? yoh: it might be better speed wise just assign dictionary
# without any shallow .copy
lcl_data = data.copy()
if copy_dsattr and len(dsattr)>0:
# deep copy
if __debug__:
debug('DS', "Deep copying dsattr %s" % `dsattr`)
lcl_dsattr = copy.deepcopy(dsattr)
else:
# shallow copy
lcl_dsattr = copy.copy(dsattr)
# has to be not private since otherwise derived methods
# would have problem accessing it and _registerAttribute
# would fail on lambda getters
self._data = lcl_data
"""What makes a dataset."""
self._dsattr = lcl_dsattr
"""Dataset attriibutes."""
# store samples (and possibly transform/reshape/retype them)
if not samples is None:
if __debug__:
if lcl_data.has_key('samples'):
debug('DS',
"`Data` dict has `samples` (%s) but there is also" \
" __init__ parameter `samples` which overrides " \
" stored in `data`" % (`lcl_data['samples'].shape`))
lcl_data['samples'] = self._shapeSamples(samples, dtype,
copy_samples)
# TODO? we might want to have the same logic for chunks and labels
# ie if no labels present -- assign arange
# MH: don't think this is necessary -- or is there a use case?
# labels
if not labels is None:
if __debug__:
if lcl_data.has_key('labels'):
debug('DS',
"`Data` dict has `labels` (%s) but there is also" +
" __init__ parameter `labels` which overrides " +
" stored in `data`" % (`lcl_data['labels']`))
if lcl_data.has_key('samples'):
lcl_data['labels'] = \
self._expandSampleAttribute(labels, 'labels')
# check if we got all required attributes
for attr in self._requiredattributes:
if not lcl_data.has_key(attr):
raise DatasetError, \
"Attribute %s is required to initialize dataset" % \
attr
nsamples = self.nsamples
# chunks
if not chunks == None:
lcl_data['chunks'] = \
self._expandSampleAttribute(chunks, 'chunks')
elif not lcl_data.has_key('chunks'):
# if no chunk information is given assume that every pattern
# is its own chunk
lcl_data['chunks'] = N.arange(nsamples)
# samples origids
if not origids is None:
# simply assign if provided
lcl_data['origids'] = origids
elif not lcl_data.has_key('origids'):
# otherwise contruct unqiue ones
lcl_data['origids'] = N.arange(len(lcl_data['labels']))
else:
# assume origids have been specified already (copy constructor
# mode) leave them as they are, e.g. to make origids survive
# selectSamples()
pass
# Initialize attributes which are registered but were not setup
for attr in self._registeredattributes:
if not lcl_data.has_key(attr):
if __debug__:
debug("DS", "Initializing attribute %s" % attr)
lcl_data[attr] = N.zeros(nsamples)
# labels_map
labels_ = N.asarray(lcl_data['labels'])
labels_map_known = lcl_dsattr.has_key('labels_map')
if labels_map is True:
# need to compose labels_map
if labels_.dtype.char == 'S': # or not labels_map_known:
# Create mapping
ulabels = list(set(labels_))
ulabels.sort()
labels_map = dict([ (x[1], x[0]) for x in enumerate(ulabels) ])
if __debug__:
debug('DS', 'Mapping for the labels computed to be %s'
% labels_map)
else:
if __debug__:
debug('DS', 'Mapping of labels was requested but labels '
'are not strings. Skipped')
labels_map = None
pass
elif labels_map is False:
labels_map = None
if isinstance(labels_map, dict):
if labels_map_known:
if __debug__:
debug('DS',
"`dsattr` dict has `labels_map` (%s) but there is also" \
" __init__ parameter `labels_map` (%s) which overrides " \
" stored in `dsattr`" % (lcl_dsattr['labels_map'], labels_map))
lcl_dsattr['labels_map'] = labels_map
# map labels if needed (if strings or was explicitely requested)
if labels_.dtype.char == 'S' or not labels_map_known:
if __debug__:
debug('DS_', "Remapping labels using mapping %s" % labels_map)
# need to remap
# !!! N.array is important here
try:
lcl_data['labels'] = N.array(
[labels_map[x] for x in lcl_data['labels']])
except KeyError, e:
raise ValueError, "Provided labels_map %s is insufficient " \
"to map all the labels. Mapping for label %s is " \
"missing" % (labels_map, e)
elif not lcl_dsattr.has_key('labels_map'):
lcl_dsattr['labels_map'] = labels_map
elif __debug__:
debug('DS_', 'Not overriding labels_map in dsattr since it has one')
if check_data:
self._checkData()
# lazy computation of unique members
#self._resetallunique('_dsattr', self._dsattr)
# Michael: we cannot do this conditional here. When selectSamples()
# removes a whole data chunk the uniquechunks values will be invalid.
# Same applies to labels of course.
if not labels is None or not chunks is None:
# for a speed up to don't go through all uniqueattributes
# when no need
lcl_dsattr['__uniquereseted'] = False
self._resetallunique(force=True)
__doc__ = enhancedDocString('Dataset', locals())
@property
def idhash(self):
"""To verify if dataset is in the same state as when smth else was done
Like if classifier was trained on the same dataset as in question"""
_data = self._data
res = idhash_(_data)
# we cannot count on the order the values in the dict will show up
# with `self._data.value()` and since idhash will be order-dependent
# we have to make it deterministic
keys = _data.keys()
keys.sort()
for k in keys:
res += idhash_(_data[k])
return res
def _resetallunique(self, force=False):
"""Set to None all unique* attributes of corresponding dictionary
"""
_dsattr = self._dsattr
if not force and _dsattr['__uniquereseted']:
return
_uniqueattributes = self._uniqueattributes
if __debug__ and "DS_" in debug.active:
debug("DS_", "Reseting all attributes %s for dataset %s"
% (_uniqueattributes,
self.summary(uniq=False, idhash=False,
stats=False, lstats=False)))
# I guess we better checked if dictname is known but...
for k in _uniqueattributes:
_dsattr[k] = None
_dsattr['__uniquereseted'] = True
def _getuniqueattr(self, attrib, dict_):
"""Provide common facility to return unique attributes
XXX `dict_` can be simply replaced now with self._dsattr
"""
# local bindings
_dsattr = self._dsattr
if not _dsattr.has_key(attrib) or _dsattr[attrib] is None:
if __debug__ and 'DS_' in debug.active:
debug("DS_", "Recomputing unique set for attrib %s within %s" %
(attrib, self.summary(uniq=False,
stats=False, lstats=False)))
# uff... might come up with better strategy to keep relevant
# attribute name
_dsattr[attrib] = N.unique( N.asanyarray(dict_[attrib[6:]]) )
assert(not _dsattr[attrib] is None)
_dsattr['__uniquereseted'] = False
return _dsattr[attrib]
def _setdataattr(self, attrib, value):
"""Provide common facility to set attributes
"""
if len(value) != self.nsamples:
raise ValueError, \
"Provided %s have %d entries while there is %d samples" % \
(attrib, len(value), self.nsamples)
self._data[attrib] = N.asarray(value)
uniqueattr = "unique" + attrib
_dsattr = self._dsattr
if _dsattr.has_key(uniqueattr):
_dsattr[uniqueattr] = None
def _getNSamplesPerAttr( self, attrib='labels' ):
"""Returns the number of samples per unique label.
"""
# local bindings
_data = self._data
# XXX hardcoded dict_=self._data.... might be in self._dsattr
uniqueattr = self._getuniqueattr(attrib="unique" + attrib,
dict_=_data)
# use dictionary to cope with arbitrary labels
result = dict(zip(uniqueattr, [ 0 ] * len(uniqueattr)))
for l in _data[attrib]:
result[l] += 1
# XXX only return values to mimic the old interface but we might want
# to return the full dict instead
# return result
return result
def _getSampleIdsByAttr(self, values, attrib="labels",
sort=True):
"""Return indecies of samples given a list of attributes
"""
if not operator.isSequenceType(values) \
or isinstance(values, basestring):
values = [ values ]
# TODO: compare to plain for loop through the labels
# on a real data example
sel = N.array([], dtype=N.int16)
_data = self._data
for value in values:
sel = N.concatenate((
sel, N.where(_data[attrib]==value)[0]))
if sort:
# place samples in the right order
sel.sort()
return sel
def idsonboundaries(self, prior=0, post=0,
attributes_to_track=['labels', 'chunks'],
affected_labels=None,
revert=False):
"""Find samples which are on the boundaries of the blocks
Such samples might need to be removed. By default (with
prior=0, post=0) ids of the first samples in a 'block' are
reported
:Parameters:
prior : int
how many samples prior to transition sample to include
post : int
how many samples post the transition sample to include
attributes_to_track : list of basestring
which attributes to track to decide on the boundary condition
affected_labels : list of basestring
for which labels to perform selection. If None - for all
revert : bool
either to revert the meaning and provide ids of samples which are found
to not to be boundary samples
"""
# local bindings
_data = self._data
labels = self.labels
nsamples = self.nsamples
lastseen = none = [None for attr in attributes_to_track]
transitions = []
for i in xrange(nsamples+1):
if i < nsamples:
current = [_data[attr][i] for attr in attributes_to_track]
else:
current = none
if lastseen != current:
# transition point
new_transitions = range(max(0, i-prior),
min(nsamples-1, i+post)+1)
if affected_labels is not None:
new_transitions = [labels[i] for i in new_transitions
if i in affected_labels]
transitions += new_transitions
lastseen = current
transitions = set(transitions)
if revert:
transitions = set(range(nsamples)).difference(transitions)
# postprocess
transitions = N.array(list(transitions))
transitions.sort()
return list(transitions)
def _shapeSamples(self, samples, dtype, copy):
"""Adapt different kinds of samples
Handle all possible input value for 'samples' and tranform
them into a 2d (samples x feature) representation.
"""
# put samples array into correct shape
# 1d arrays or simple sequences are assumed to be a single pattern
if (not isinstance(samples, N.ndarray)):
# it is safe to provide dtype which defaults to None,
# when N would choose appropriate dtype automagically
samples = N.array(samples, ndmin=2, dtype=dtype, copy=copy)
else:
if samples.ndim < 2 \
or (not dtype is None and dtype != samples.dtype):
if dtype is None:
dtype = samples.dtype
samples = N.array(samples, ndmin=2, dtype=dtype, copy=copy)
elif copy:
samples = samples.copy()
# only samples x features matrices are supported
if len(samples.shape) > 2:
raise DatasetError, "Only (samples x features) -> 2d sample " \
+ "are supported (got %s shape of samples)." \
% (`samples.shape`) \
+" Consider MappedDataset if applicable."
return samples
def _checkData(self):
"""Checks `_data` members to have the same # of samples.
"""
#
# XXX: Maybe just run this under __debug__ and remove the `check_data`
# from the constructor, which is too complicated anyway?
#
# local bindings
nsamples = self.nsamples
_data = self._data
for k, v in _data.iteritems():
if not len(v) == nsamples:
raise DatasetError, \
"Length of sample attribute '%s' [%i] does not " \
"match the number of samples in the dataset [%i]." \
% (k, len(v), nsamples)
# check for unique origids
uniques = N.unique(_data['origids'])
uniques.sort()
# need to copy to prevent sorting the original array
sorted_ids = _data['origids'].copy()
sorted_ids.sort()
if not (uniques == sorted_ids).all():
raise DatasetError, "Samples IDs are not unique."
# Check if labels as not literal
if N.asanyarray(_data['labels'].dtype.char == 'S'):
warning('Labels for dataset %s are literal, should be numeric. '
'You might like to use labels_map argument.' % self)
def _expandSampleAttribute(self, attr, attr_name):
"""If a sample attribute is given as a scalar expand/repeat it to a
length matching the number of samples in the dataset.
"""
try:
# if we are initializing with a single string -- we should
# treat it as a single label
if isinstance(attr, basestring):
raise TypeError
if len(attr) != self.nsamples:
raise DatasetError, \
"Length of sample attribute '%s' [%d]" \
% (attr_name, len(attr)) \
+ " has to match the number of samples" \
+ " [%d]." % self.nsamples
# store the sequence as array
return N.array(attr)
except TypeError:
# make sequence of identical value matching the number of
# samples
return N.repeat(attr, self.nsamples)
@classmethod
def _registerAttribute(cls, key, dictname="_data", abbr=None, hasunique=False):
"""Register an attribute for any Dataset class.
Creates property assigning getters/setters depending on the
availability of corresponding _get, _set functions.
"""
classdict = cls.__dict__
if not classdict.has_key(key):
if __debug__:
debug("DS", "Registering new attribute %s" % key)
# define get function and use corresponding
# _getATTR if such defined
getter = '_get%s' % key
if classdict.has_key(getter):
getter = '%s.%s' % (cls.__name__, getter)
else:
getter = "lambda x: x.%s['%s']" % (dictname, key)
# define set function and use corresponding
# _setATTR if such defined
setter = '_set%s' % key
if classdict.has_key(setter):
setter = '%s.%s' % (cls.__name__, setter)
elif dictname=="_data":
setter = "lambda self,x: self._setdataattr" + \
"(attrib='%s', value=x)" % (key)
else:
setter = None
if __debug__:
debug("DS", "Registering new property %s.%s" %
(cls.__name__, key))
exec "%s.%s = property(fget=%s, fset=%s)" % \
(cls.__name__, key, getter, setter)
if abbr is not None:
exec "%s.%s = property(fget=%s, fset=%s)" % \
(cls.__name__, abbr, getter, setter)
if hasunique:
uniquekey = "unique%s" % key
getter = '_get%s' % uniquekey
if classdict.has_key(getter):
getter = '%s.%s' % (cls.__name__, getter)
else:
getter = "lambda x: x._getuniqueattr" + \
"(attrib='%s', dict_=x.%s)" % (uniquekey, dictname)
if __debug__:
debug("DS", "Registering new property %s.%s" %
(cls.__name__, uniquekey))
exec "%s.%s = property(fget=%s)" % \
(cls.__name__, uniquekey, getter)
if abbr is not None:
exec "%s.U%s = property(fget=%s)" % \
(cls.__name__, abbr, getter)
# create samplesper<ATTR> properties
sampleskey = "samplesper%s" % key[:-1] # remove ending 's' XXX
if __debug__:
debug("DS", "Registering new property %s.%s" %
(cls.__name__, sampleskey))
exec "%s.%s = property(fget=%s)" % \
(cls.__name__, sampleskey,
"lambda x: x._getNSamplesPerAttr(attrib='%s')" % key)
cls._uniqueattributes.append(uniquekey)
# create idsby<ATTR> properties
sampleskey = "idsby%s" % key # remove ending 's' XXX
if __debug__:
debug("DS", "Registering new property %s.%s" %
(cls.__name__, sampleskey))
exec "%s.%s = %s" % (cls.__name__, sampleskey,
"lambda self, x: " +
"self._getSampleIdsByAttr(x,attrib='%s')" % key)
cls._uniqueattributes.append(uniquekey)
cls._registeredattributes.append(key)
elif __debug__:
warning('Trying to reregister attribute `%s`. For now ' % key +
'such capability is not present')
def __str__(self):
"""String summary over the object
"""
try:
ssummary = self.summary(uniq=True,
idhash=__debug__ and ('DS_ID' in debug.active),
stats=__debug__ and ('DS_STATS' in debug.active),
lstats=__debug__ and ('DS_STATS' in debug.active),
)
except (AttributeError, KeyError), e:
# __str__ or __repr__ could have been requested before actual
# instance is populated, e.g. by tracebacks of pdb/pydb.
# ??? this case might be generic enough to allow for common
# decorator around plentiful of __str__ and __repr__s
ssummary = str(e)
return ssummary
def __repr__(self):
return "<%s>" % str(self)
def summary(self, uniq=True, stats=True, idhash=False, lstats=True,
maxc=30, maxl=20):
"""String summary over the object
:Parameters:
uniq : bool
Include summary over data attributes which have unique
idhash : bool
Include idhash value for dataset and samples
stats : bool
Include some basic statistics (mean, std, var) over dataset samples
lstats : bool
Include statistics on chunks/labels
maxc : int
Maximal number of chunks when provide details on labels/chunks
maxl : int
Maximal number of labels when provide details on labels/chunks
"""
# local bindings
samples = self.samples
_data = self._data
_dsattr = self._dsattr
if idhash:
idhash_ds = "{%s}" % self.idhash
idhash_samples = "{%s}" % idhash_(samples)
else:
idhash_ds = ""
idhash_samples = ""
s = """Dataset %s/ %s %d%s x %d""" % \
(idhash_ds, samples.dtype,
self.nsamples, idhash_samples, self.nfeatures)
ssep = (' ', '\n')[lstats]
if uniq:
s += "%suniq:" % ssep
for uattr in _dsattr.keys():
if not uattr.startswith("unique"):
continue
attr = uattr[6:]
try:
value = self._getuniqueattr(attrib=uattr,
dict_=_data)
s += " %d %s" % (len(value), attr)
except:
pass
if isinstance(self.labels_map, dict):
s += ' labels_mapped'
if stats:
# TODO -- avg per chunk?
# XXX We might like to use scipy.stats.describe to get
# quick summary statistics (mean/range/skewness/kurtosis)
if self.nfeatures:
s += "%sstats: mean=%g std=%g var=%g min=%g max=%g\n" % \
(ssep, N.mean(samples), N.std(samples),
N.var(samples), N.min(samples), N.max(samples))
else:
s += "%sstats: dataset has no features\n" % ssep
if lstats:
s += self.summary_labels(maxc=maxc, maxl=maxl)
return s
def summary_labels(self, maxc=30, maxl=20):
"""Provide summary statistics over the labels and chunks
:Parameters:
maxc : int
Maximal number of chunks when provide details
maxl : int
Maximal number of labels when provide details
"""
# We better avoid bound function since if people only
# imported Dataset without miscfx it would fail
from mvpa.datasets.miscfx import getSamplesPerChunkLabel
spcl = getSamplesPerChunkLabel(self)
# XXX couldn't they be unordered?
ul = self.uniquelabels.tolist()
uc = self.uniquechunks.tolist()
s = ""
if len(ul) < maxl and len(uc) < maxc:
s += "\nCounts of labels in each chunk:"
# only in a resonable case do printing
table = [[' chunks\labels'] + ul]
table += [[''] + ['---'] * len(ul)]
for c, counts in zip(uc, spcl):
table.append([ str(c) ] + counts.tolist())
s += '\n' + table2string(table)
else:
s += "No details due to large number of labels or chunks. " \
"Increase maxc and maxl if desired"
labels_map = self.labels_map
if isinstance(labels_map, dict):
s += "\nOriginal labels were mapped using following mapping:"
s += '\n\t'+'\n\t'.join([':\t'.join(map(str, x))
for x in labels_map.items()]) + '\n'
def cl_stats(axis, u, name1, name2):
""" Compute statistics per label
"""
stats = {'min': N.min(spcl, axis=axis),
'max': N.max(spcl, axis=axis),
'mean': N.mean(spcl, axis=axis),
'std': N.std(spcl, axis=axis),
'#%ss' % name2: N.sum(spcl>0, axis=axis)}
entries = [' ' + name1, 'mean', 'std', 'min', 'max', '#%ss' % name2]
table = [ entries ]
for i, l in enumerate(u):
d = {' ' + name1 : l}
d.update(dict([ (k, stats[k][i]) for k in stats.keys()]))
table.append( [ ('%.3g', '%s')[isinstance(d[e], basestring)]
% d[e] for e in entries] )
return '\nSummary per %s across %ss\n' % (name1, name2) \
+ table2string(table)
if len(ul) < maxl:
s += cl_stats(0, ul, 'label', 'chunk')
if len(uc) < maxc:
s += cl_stats(1, uc, 'chunk', 'label')
return s
def __iadd__(self, other):
"""Merge the samples of one Dataset object to another (in-place).
No dataset attributes, besides labels_map, will be merged!
Additionally, a new set of unique `origids` will be generated.
"""
# local bindings
_data = self._data
other_data = other._data
if not self.nfeatures == other.nfeatures:
raise DatasetError, "Cannot add Dataset, because the number of " \
"feature do not match."
# take care about labels_map and labels
slm = self.labels_map
olm = other.labels_map
if N.logical_xor(slm is None, olm is None):
raise ValueError, "Cannot add datasets where only one of them " \
"has labels map assigned. If needed -- implement it"
# concatenate all sample attributes
for k,v in _data.iteritems():
if k == 'origids':
# special case samples origids: for now just regenerate unique
# ones could also check if concatenation is unique, but it
# would be costly performance-wise
_data[k] = N.arange(len(v) + len(other_data[k]))
elif k == 'labels' and slm is not None:
# special care about labels if mapping was in effect,
# we need to append 2nd map to the first one and
# relabel 2nd dataset
nlm = slm.copy()
# figure out maximal numerical label used now
nextid = N.sort(nlm.values())[-1] + 1
olabels = other.labels
olabels_remap = {}
for ol, olnum in olm.iteritems():
if not nlm.has_key(ol):
# check if we can preserve old numberic label
# if not -- assign some new one not yet present
# in any dataset
if olnum in nlm.values():
nextid = N.sort(nlm.values() + olm.values())[-1] + 1
else:
nextid = olnum
olabels_remap[olnum] = nextid
nlm[ol] = nextid
nextid += 1
else:
olabels_remap[olnum] = nlm[ol]
olabels = [olabels_remap[x] for x in olabels]
# finally compose new labels
_data['labels'] = N.concatenate((v, olabels), axis=0)
# and reassign new mapping
self._dsattr['labels_map'] = nlm
if __debug__:
# check if we are not dealing with colliding
# mapping, since it is problematic and might lead
# to various complications
if (len(set(slm.keys())) != len(set(slm.values()))) or \
(len(set(olm.keys())) != len(set(olm.values()))):
warning("Adding datasets where multiple labels "
"mapped to the same ID is not recommended. "
"Please check the outcome. Original mappings "
"were %s and %s. Resultant is %s"
% (slm, olm, nlm))
else:
_data[k] = N.concatenate((v, other_data[k]), axis=0)
# might be more sophisticated but for now just reset -- it is safer ;)
self._resetallunique()
return self
def __add__( self, other ):
"""Merge the samples two Dataset objects.
All data of both datasets is copied, concatenated and a new Dataset is
returned.
NOTE: This can be a costly operation (both memory and time). If
performance is important consider the '+=' operator.
"""
# create a new object of the same type it is now and NOT only Dataset
out = super(Dataset, self).__new__(self.__class__)
# now init it: to make it work all Dataset contructors have to accept
# Class(data=Dict, dsattr=Dict)
out.__init__(data=self._data,
dsattr=self._dsattr,
copy_samples=True,
copy_data=True,
copy_dsattr=True)
out += other
return out
def copy(self, deep=True):
"""Create a copy (clone) of the dataset, by fully copying current one
:Keywords:
deep : bool
deep flag is provided to __init__ for
copy_{samples,data,dsattr}. By default full copy is done.
"""
# create a new object of the same type it is now and NOT only Dataset
out = super(Dataset, self).__new__(self.__class__)
# now init it: to make it work all Dataset contructors have to accept
# Class(data=Dict, dsattr=Dict)
out.__init__(data=self._data,
dsattr=self._dsattr,
copy_samples=True,
copy_data=True,
copy_dsattr=True)
return out
def selectFeatures(self, ids=None, sort=True, groups=None):
"""Select a number of features from the current set.
:Parameters:
ids
iterable container to select ids
sort : bool
if to sort Ids. Order matters and `selectFeatures` assumes
incremental order. If not such, in non-optimized code
selectFeatures would verify the order and sort
Returns a new Dataset object with a copy of corresponding features
from the original samples array.
WARNING: The order of ids determines the order of features in
the returned dataset. This might be useful sometimes, but can
also cause major headaches! Order would is verified when
running in non-optimized code (if __debug__)
"""
if ids is None and groups is None:
raise ValueError, "No feature selection specified."
# start with empty list if no ids where specified (so just groups)
if ids is None:
ids = []
if not groups is None:
if not self._dsattr.has_key('featuregroups'):
raise RuntimeError, \
"Dataset has no feature grouping information."
for g in groups:
ids += (self._dsattr['featuregroups'] == g).nonzero()[0].tolist()
# XXX set sort default to True, now sorting has to be explicitely
# disabled and warning is not necessary anymore
if sort:
ids = copy.deepcopy(ids)
ids.sort()
elif __debug__ and 'CHECK_DS_SORTED' in debug.active:
from mvpa.misc.support import isSorted
if not isSorted(ids):
warning("IDs for selectFeatures must be provided " +
"in sorted order, otherwise major headache might occur")
# shallow-copy all stuff from current data dict
new_data = self._data.copy()
# assign the selected features -- data is still shared with
# current dataset
new_data['samples'] = self._data['samples'][:, ids]
# apply selection to feature groups as well
if self._dsattr.has_key('featuregroups'):
new_dsattr = self._dsattr.copy()
new_dsattr['featuregroups'] = self._dsattr['featuregroups'][ids]
else:
new_dsattr = self._dsattr
# create a new object of the same type it is now and NOT only Dataset
dataset = super(Dataset, self).__new__(self.__class__)
# now init it: to make it work all Dataset contructors have to accept
# Class(data=Dict, dsattr=Dict)
dataset.__init__(data=new_data,
dsattr=new_dsattr,
check_data=False,
copy_samples=False,
copy_data=False,
copy_dsattr=False
)
return dataset
def applyMapper(self, featuresmapper=None, samplesmapper=None,
train=True):
"""Obtain new dataset by applying mappers over features and/or samples.
While featuresmappers leave the sample attributes information
unchanged, as the number of samples in the dataset is invariant,
samplesmappers are also applied to the samples attributes themselves!
Applying a featuresmapper will destroy any feature grouping information.
:Parameters:
featuresmapper : Mapper
`Mapper` to somehow transform each sample's features
samplesmapper : Mapper
`Mapper` to transform each feature across samples
train : bool
Flag whether to train the mapper with this dataset before applying
it.
TODO: selectFeatures is pretty much
applyMapper(featuresmapper=MaskMapper(...))
"""
# shallow-copy all stuff from current data dict
new_data = self._data.copy()
# apply mappers
if samplesmapper:
if __debug__:
debug("DS", "Training samplesmapper %s" % `samplesmapper`)
samplesmapper.train(self)
if __debug__:
debug("DS", "Applying samplesmapper %s" % `samplesmapper` +
" to samples of dataset `%s`" % `self`)
# get rid of existing 'origids' as they are not valid anymore and
# applying a mapper to them is not really meaningful
if new_data.has_key('origids'):
del(new_data['origids'])
# apply mapper to all sample-wise data in dataset
for k in new_data.keys():
new_data[k] = samplesmapper.forward(self._data[k])
# feature mapping might affect dataset attributes
# XXX: might be obsolete when proper feature attributes are implemented
new_dsattr = self._dsattr
if featuresmapper:
if __debug__:
debug("DS", "Training featuresmapper %s" % `featuresmapper`)
featuresmapper.train(self)
if __debug__:
debug("DS", "Applying featuresmapper %s" % `featuresmapper` +
" to samples of dataset `%s`" % `self`)
new_data['samples'] = featuresmapper.forward(self._data['samples'])
# remove feature grouping, who knows what the mapper did to the
# features
if self._dsattr.has_key('featuregroups'):
new_dsattr = self._dsattr.copy()
del(new_dsattr['featuregroups'])
else:
new_dsattr = self._dsattr
# create a new object of the same type it is now and NOT only Dataset
dataset = super(Dataset, self).__new__(self.__class__)
# now init it: to make it work all Dataset contructors have to accept
# Class(data=Dict, dsattr=Dict)
dataset.__init__(data=new_data,
dsattr=new_dsattr,
check_data=False,
copy_samples=False,
copy_data=False,
copy_dsattr=False
)
# samples attributes might have changed after applying samplesmapper
if samplesmapper:
dataset._resetallunique(force=True)
return dataset
def selectSamples(self, ids):
"""Choose a subset of samples defined by samples IDs.
Returns a new dataset object containing the selected sample
subset.
TODO: yoh, we might need to sort the mask if the mask is a
list of ids and is not ordered. Clarify with Michael what is
our intent here!
"""
# without having a sequence a index the masked sample array would
# loose its 2d layout
if not operator.isSequenceType( ids ):
ids = [ids]
# TODO: Reconsider crafting a slice if it can be done to don't copy
# the data
#try:
# minmask = min(mask)
# maxmask = max(mask)
#except:
# minmask = min(map(int,mask))
# maxmask = max(map(int,mask))
# lets see if we could get it done with cheap view/slice
#(minmask, maxmask) != (0, 1) and \
#if len(mask) > 2 and \
# N.array([N.arange(minmask, maxmask+1) == N.array(mask)]).all():
# slice_ = slice(minmask, maxmask+1)
# if __debug__:
# debug("DS", "We can and do convert mask %s into splice %s" %
# (mask, slice_))
# mask = slice_
# mask all sample attributes
data = {}
for k, v in self._data.iteritems():
data[k] = v[ids, ]
# create a new object of the same type it is now and NOT onyl Dataset
dataset = super(Dataset, self).__new__(self.__class__)
# now init it: to make it work all Dataset contructors have to accept
# Class(data=Dict, dsattr=Dict)
dataset.__init__(data=data,
dsattr=self._dsattr,
check_data=False,
copy_samples=False,
copy_data=False,
copy_dsattr=False)
dataset._resetallunique(force=True)
return dataset
def index(self, *args, **kwargs):
"""Universal indexer to obtain indexes of interesting samples/features.
See .select() for more information
:Return: tuple of (samples indexes, features indexes). Each
item could be also None, if no selection on samples or
features was requested (to discriminate between no selected
items, and no selections)
"""
s_indx = [] # selections for samples
f_indx = [] # selections for features
return_dataset = kwargs.pop('return_dataset', False)
largs = len(args)
args = list(args) # so we could override
# Figure out number of positional
largs_nonstring = 0
# need to go with index since we might need to override internally
for i in xrange(largs):
l = args[i]
if isinstance(l, basestring):
if l.lower() == 'all':
# override with a slice
args[i] = slice(None)
else:
break
largs_nonstring += 1
if largs_nonstring >= 1:
s_indx.append(args[0])
if __debug__ and 'CHECK_DS_SELECT' in debug.active:
_validate_indexes_uniq_sorted(args[0], 'select', 'samples')
if largs_nonstring == 2:
f_indx.append(args[1])
if __debug__ and 'CHECK_DS_SELECT' in debug.active:
_validate_indexes_uniq_sorted(args[1], 'select', 'features')
elif largs_nonstring > 2:
raise ValueError, "Only two positional arguments are allowed" \
". 1st for samples, 2nd for features"
# process left positional arguments which must encode selections like
# ('labels', [1,2,3])
if (largs - largs_nonstring) % 2 != 0:
raise ValueError, "Positional selections must come in pairs:" \
" e.g. ('labels', [1,2,3])"
for i in xrange(largs_nonstring, largs, 2):
k, v = args[i:i+2]
kwargs[k] = v
# process keyword parameters
data_ = self._data
for k, v in kwargs.iteritems():
if k == 'samples':
s_indx.append(v)
elif k == 'features':
f_indx.append(v)
elif data_.has_key(k):
# so it is an attribute for samples
# XXX may be do it not only if __debug__
if __debug__: # and 'CHECK_DS_SELECT' in debug.active:
if not N.any([isinstance(v, cls) for cls in
[list, tuple, slice, int]]):
raise ValueError, "Trying to specify selection for %s " \
"based on unsupported '%s'" % (k, v)
s_indx.append(self._getSampleIdsByAttr(v, attrib=k, sort=False))
else:
raise ValueError, 'Keyword "%s" is not known, thus' \
'select() failed' % k
def combine_indexes(indx, nelements):
"""Helper function: intersect selections given in indx
:Parameters:
indxs : list of lists or slices
selections of elements
nelements : int
number of elements total for deriving indexes from slices
"""
indx_sel = None # pure list of ids for selection
for s in indx:
if isinstance(s, slice) or \
isinstance(s, N.ndarray) and s.dtype==bool:
# XXX there might be a better way than reconstructing the full
# index list. Also we are loosing ability to do simlpe slicing,
# ie w.o making a copy of the selected data
all_indexes = N.arange(nelements)
s = all_indexes[s]
elif not operator.isSequenceType(s):
s = [ s ]
if indx_sel is None:
indx_sel = set(s)
else:
# To be consistent
#if not isinstance(indx_sel, set):
# indx_sel = set(indx_sel)
indx_sel = indx_sel.intersection(s)
# if we got set -- convert
if isinstance(indx_sel, set):
indx_sel = list(indx_sel)
# sort for the sake of sanity
indx_sel.sort()
return indx_sel
# Select samples
if len(s_indx) == 1 and isinstance(s_indx[0], slice) \
and s_indx[0] == slice(None):
# so no actual selection -- full slice
s_indx = s_indx[0]
else:
# else - get indexes
if len(s_indx) == 0:
s_indx = None
else:
s_indx = combine_indexes(s_indx, self.nsamples)
# Select features
if len(f_indx):
f_indx = combine_indexes(f_indx, self.nfeatures)
else:
f_indx = None
return s_indx, f_indx
def select(self, *args, **kwargs):
"""Universal selector
WARNING: if you need to select duplicate samples
(e.g. samples=[5,5]) or order of selected samples of features
is important and has to be not ordered (e.g. samples=[3,2,1]),
please use selectFeatures or selectSamples functions directly
Examples:
Mimique plain selectSamples::
dataset.select([1,2,3])
dataset[[1,2,3]]
Mimique plain selectFeatures::
dataset.select(slice(None), [1,2,3])
dataset.select('all', [1,2,3])
dataset[:, [1,2,3]]
Mixed (select features and samples)::
dataset.select([1,2,3], [1, 2])
dataset[[1,2,3], [1, 2]]
Select samples matching some attributes::
dataset.select(labels=[1,2], chunks=[2,4])
dataset.select('labels', [1,2], 'chunks', [2,4])
dataset['labels', [1,2], 'chunks', [2,4]]
Mixed -- out of first 100 samples, select only those with
labels 1 or 2 and belonging to chunks 2 or 4, and select
features 2 and 3::
dataset.select(slice(0,100), [2,3], labels=[1,2], chunks=[2,4])
dataset[:100, [2,3], 'labels', [1,2], 'chunks', [2,4]]
"""
s_indx, f_indx = self.index(*args, **kwargs)
# Select samples
if s_indx == slice(None):
# so no actual selection was requested among samples.
# thus proceed with original dataset
if __debug__:
debug('DS', 'in select() not selecting samples')
ds = self
else:
# else do selection
if __debug__:
debug('DS', 'in select() selecting samples given selections'
+ str(s_indx))
ds = self.selectSamples(s_indx)
# Select features
if f_indx is not None:
if __debug__:
debug('DS', 'in select() selecting features given selections'
+ str(f_indx))
ds = ds.selectFeatures(f_indx)
return ds
def where(self, *args, **kwargs):
"""Obtain indexes of interesting samples/features. See select() for more information
XXX somewhat obsoletes idsby...
"""
s_indx, f_indx = self.index(*args, **kwargs)
if s_indx is not None and f_indx is not None:
return s_indx, f_indx
elif s_indx is not None:
return s_indx
else:
return f_indx
def __getitem__(self, *args):
"""Convinience dataset parts selection
See select for more information
"""
# for cases like ['labels', 1]
if len(args) == 1 and isinstance(args[0], tuple):
args = args[0]
args_, args = args, ()
for a in args_:
if isinstance(a, slice) and \
isinstance(a.start, basestring):
# for the constructs like ['labels':[1,2]]
if a.stop is None or a.step is not None:
raise ValueError, \
"Selection must look like ['chunks':[2,3]]"
args += (a.start, a.stop)
else:
args += (a,)
return self.select(*args)
def permuteLabels(self, status, perchunk=True, assure_permute=False):
"""Permute the labels.
TODO: rename status into something closer in semantics.
:Parameters:
status : bool
Calling this method with set to True, the labels are
permuted among all samples. If 'status' is False the
original labels are restored.
perchunk : bool
If True permutation is limited to samples sharing the same
chunk value. Therefore only the association of a certain
sample with a label is permuted while keeping the absolute
number of occurences of each label value within a certain
chunk constant.
assure_permute : bool
If True, assures that labels are permutted, ie any one is
different from the original one
"""
# local bindings
_data = self._data
if len(self.uniquelabels)<2:
raise RuntimeError, \
"Call to permuteLabels is bogus since there is insuficient" \
" number of labels: %s" % self.uniquelabels
if not status:
# restore originals
if _data.get('origlabels', None) is None:
raise RuntimeError, 'Cannot restore labels. ' \
'permuteLabels() has never been ' \
'called with status == True.'
self.labels = _data['origlabels']
_data.pop('origlabels')
else:
# store orig labels, but only if not yet done, otherwise multiple
# calls with status == True will destroy the original labels
if not _data.has_key('origlabels') \
or _data['origlabels'] == None:
# bind old labels to origlabels
_data['origlabels'] = _data['labels']
# copy labels
_data['labels'] = copy.copy(_data['labels'])
labels = _data['labels']
# now scramble
if perchunk:
for o in self.uniquechunks:
labels[self.chunks == o] = \
N.random.permutation(labels[self.chunks == o])
else:
labels = N.random.permutation(labels)
self.labels = labels
if assure_permute:
if not (_data['labels'] != _data['origlabels']).any():
if not (assure_permute is True):
if assure_permute == 1:
raise RuntimeError, \
"Cannot assure permutation of labels %s for " \
"some reason with chunks %s and while " \
"perchunk=%s . Should not happen" % \
(self.labels, self.chunks, perchunk)
else:
assure_permute = 11 # make 10 attempts
if __debug__:
debug("DS", "Recalling permute to assure different labels")
self.permuteLabels(status, perchunk=perchunk,
assure_permute=assure_permute-1)
def getRandomSamples(self, nperlabel):
"""Select a random set of samples.
If 'nperlabel' is an integer value, the specified number of samples is
randomly choosen from the group of samples sharing a unique label
value ( total number of selected samples: nperlabel x len(uniquelabels).
If 'nperlabel' is a list which's length has to match the number of
unique label values. In this case 'nperlabel' specifies the number of
samples that shall be selected from the samples with the corresponding
label.
The method returns a Dataset object containing the selected
samples.
"""
# if interger is given take this value for all classes
if isinstance(nperlabel, int):
nperlabel = [ nperlabel for i in self.uniquelabels ]
sample = []
# for each available class
labels = self.labels
for i, r in enumerate(self.uniquelabels):
# get the list of pattern ids for this class
sample += random.sample( (labels == r).nonzero()[0],
nperlabel[i] )
return self.selectSamples( sample )
# def _setchunks(self, chunks):
# """Sets chunks and recomputes uniquechunks
# """
# self._data['chunks'] = N.array(chunks)
# self._dsattr['uniquechunks'] = None # None!since we might not need them
def getNSamples( self ):
"""Currently available number of patterns.
"""
return self._data['samples'].shape[0]
def getNFeatures( self ):
"""Number of features per pattern.
"""
return self._data['samples'].shape[1]
def getLabelsMap(self):
"""Stored labels map (if any)
"""
return self._dsattr.get('labels_map', None)
def setLabelsMap(self, lm):
"""Set labels map.
Checks for the validity of the mapping -- values should cover
all existing labels in the dataset
"""
values = set(lm.values())
labels = set(self.uniquelabels)
if not values.issuperset(labels):
raise ValueError, \
"Provided mapping %s has some existing labels (out of %s) " \
"missing from mapping" % (list(values), list(labels))
self._dsattr['labels_map'] = lm
def setSamplesDType(self, dtype):
"""Set the data type of the samples array.
"""
# local bindings
_data = self._data
if _data['samples'].dtype != dtype:
_data['samples'] = _data['samples'].astype(dtype)
def defineFeatureGroups(self, definition):
"""Assign `definition` to featuregroups
XXX Feature-groups was not finished to be useful
"""
if not len(definition) == self.nfeatures:
raise ValueError, \
"Length of feature group definition %i " \
"does not match the number of features %i " \
% (len(definition), self.nfeatures)
self._dsattr['featuregroups'] = N.array(definition)
def convertFeatureIds2FeatureMask(self, ids):
"""Returns a boolean mask with all features in `ids` selected.
:Parameters:
ids: list or 1d array
To be selected features ids.
:Returns:
ndarray: dtype='bool'
All selected features are set to True; False otherwise.
"""
fmask = N.repeat(False, self.nfeatures)
fmask[ids] = True
return fmask
def convertFeatureMask2FeatureIds(self, mask):
"""Returns feature ids corresponding to non-zero elements in the mask.
:Parameters:
mask: 1d ndarray
Feature mask.
:Returns:
ndarray: integer
Ids of non-zero (non-False) mask elements.
"""
return mask.nonzero()[0]
@staticmethod
def _checkCopyConstructorArgs(**kwargs):
"""Common sanity check for Dataset copy constructor calls."""
# check if we have samples (somwhere)
samples = None
if kwargs.has_key('samples'):
samples = kwargs['samples']
if samples is None and kwargs.has_key('data') \
and kwargs['data'].has_key('samples'):
samples = kwargs['data']['samples']
if samples is None:
raise DatasetError, \
"`samples` must be provided to copy constructor call."
if not len(samples.shape) == 2:
raise DatasetError, \
"samples must be in 2D shape in copy constructor call."
# read-only class properties
nsamples = property( fget=getNSamples )
nfeatures = property( fget=getNFeatures )
labels_map = property( fget=getLabelsMap, fset=setLabelsMap )
def datasetmethod(func):
"""Decorator to easily bind functions to a Dataset class
"""
if __debug__:
debug("DS_", "Binding function %s to Dataset class" % func.func_name)
# Bind the function
setattr(Dataset, func.func_name, func)
# return the original one
return func
# Following attributes adherent to the basic dataset
Dataset._registerAttribute("samples", "_data", abbr='S', hasunique=False)
Dataset._registerAttribute("labels", "_data", abbr='L', hasunique=True)
Dataset._registerAttribute("chunks", "_data", abbr='C', hasunique=True)
# samples ids (already unique by definition)
Dataset._registerAttribute("origids", "_data", abbr='I', hasunique=False)
# XXX This is the place to redo the Dataset base class in a more powerful, yet
# simpler way. The basic goal is to allow for all kinds of attributes:
#
# 1) Samples attributes (per-sample full)
# 2) Features attributes (per-feature stuff)
#
# 3) Dataset attributes (per-dataset stuff)
#
# Use cases:
#
# 1) labels and chunks -- goal: it should be possible to have multivariate
# labels, e.g. to specify targets for a neural network output layer
#
# 2) feature binding/grouping -- goal: easily define ROIs in datasets, or
# group/mark various types of feature so they could be selected or
# discarded all together
#
# 3) Mappers, or chains of them (this should be possible already, but could
# be better integrated to make applyMapper() obsolete).
#
#
# Perform distortion correction on __init__(). The copy contructor
# implementation should move into a separate classmethod.
#
# Think about implementing the current 'clever' attributes in terms of one-time
# properties as suggested by Fernando on nipy-devel.
# ...
from mvpa.misc.state import ClassWithCollections, Collection
from mvpa.misc.attributes import SampleAttribute, FeatureAttribute, \
DatasetAttribute
# Remaining public interface of Dataset
class _Dataset(ClassWithCollections):
"""The successor of Dataset.
"""
# placeholder for all three basic collections of a Dataset
# put here to be able to check whether the AttributesCollector already
# instanciated a particular collection
# XXX maybe it should not do this at all for Dataset
sa = None
fa = None
dsa = None
# storage of samples in a plain NumPy array for fast access
samples = None
def __init__(self, samples, sa=None, fa=None, dsa=None):
"""
This is the generic internal constructor. Its main task is to allow
for a maximum level of customization during dataset construction,
including fast copy construction.
Parameters
----------
samples : ndarray
Data samples.
sa : Collection
Samples attributes collection.
fa : Collection
Features attributes collection.
dsa : Collection
Dataset attributes collection.
"""
# init base class
ClassWithCollections.__init__(self)
# Internal constructor -- users focus on init* Methods
# Every dataset needs data (aka samples), completely data-driven
# analyses might not even need labels, so this is the only mandatory
# argument
# XXX add checks
self.samples = samples
# Everything else in a dataset (except for samples) is organized in
# collections
# copy attributes from source collections (scol) into target
# collections (tcol)
for scol, tcol in ((sa, self.sa),
(fa, self.fa),
(dsa, self.dsa)):
# make sure we have the target collection
if tcol is None:
# XXX maybe use different classes for the collections
# but currently no reason to do so
tcol = Collection(owner=self)
# transfer the attributes
if not scol is None:
for name, attr in scol.items.iteritems():
# this will also update the owner of the attribute
# XXX discuss the implications of always copying
tcol.add(copy.copy(attr))
@classmethod
def initSimple(klass, samples, labels, chunks):
# use Numpy convention
"""
One line summary.
Long description.
Parameters
----------
samples : ndarray
The two-dimensional samples matrix.
labels : ndarray
chunks : ndarray
Returns
-------
blah blah
Notes
-----
blah blah
See Also
--------
blah blah
Examples
--------
blah blah
"""
# Demo user contructor
# compile the necessary samples attributes collection
labels_ = SampleAttribute(name='labels')
labels_.value = labels
chunks_ = SampleAttribute(name='chunks')
chunks_.value = chunks
# feels strange that one has to give the name again
# XXX why does items have to be a dict when each samples
# attr already knows its name
sa = Collection(items={'labels': labels_, 'chunks': chunks_})
# common checks should go into __init__
return klass(samples, sa=sa)
def getNSamples( self ):
"""Currently available number of patterns.
"""
return self.samples.shape[0]
def getNFeatures( self ):
"""Number of features per pattern.
"""
return self.samples.shape[1]
#
# @property
# def idhash(self):
# pass
#
#
# def idsonboundaries(self, prior=0, post=0,
# attributes_to_track=['labels', 'chunks'],
# affected_labels=None,
# revert=False):
# pass
#
#
# def summary(self, uniq=True, stats=True, idhash=False, lstats=True,
# maxc=30, maxl=20):
# pass
#
#
# def summary_labels(self, maxc=30, maxl=20):
# pass
#
#
# def __iadd__(self, other):
# pass
#
#
# def __add__( self, other ):
# pass
#
#
# def copy(self):
# pass
#
#
# def selectFeatures(self, ids=None, sort=True, groups=None):
# pass
#
#
# def applyMapper(self, featuresmapper=None, samplesmapper=None,
# train=True):
# pass
#
#
# def selectSamples(self, ids):
# pass
#
#
# def index(self, *args, **kwargs):
# pass
#
#
# def select(self, *args, **kwargs):
# pass
#
#
# def where(self, *args, **kwargs):
# pass
#
#
# def __getitem__(self, *args):
# pass
#
#
# def permuteLabels(self, status, perchunk=True, assure_permute=False):
# pass
#
#
# def getRandomSamples(self, nperlabel):
# pass
#
#
# def getLabelsMap(self):
# pass
#
#
# def setLabelsMap(self, lm):
# pass
#
#
# def setSamplesDType(self, dtype):
# pass
#
#
# def defineFeatureGroups(self, definition):
# pass
#
#
# def convertFeatureIds2FeatureMask(self, ids):
# pass
#
#
# def convertFeatureMask2FeatureIds(self, mask):
# pass
|