some new features

.venv/lib/python3.12/site-packages/statsmodels/datasets/cancer/__init__.py

@@ -0,0 +1,5 @@
+__all__ = ["load", "load_pandas",
+           "COPYRIGHT", "TITLE", "SOURCE", "DESCRSHORT", "DESCRLONG", "NOTE"]
+from .data import (
+    load, load_pandas,
+    COPYRIGHT, TITLE, SOURCE, DESCRSHORT, DESCRLONG, NOTE)

.venv/lib/python3.12/site-packages/statsmodels/datasets/cancer/cancer.csv

@@ -0,0 +1,302 @@
+cancer,population
+1,445
+0,559
+3,677
+4,681
+3,746
+4,869
+1,950
+5,976
+5,1096
+5,1098
+5,1114
+7,1125
+5,1236
+6,1285
+3,1291
+3,1318
+2,1323
+8,1327
+9,1438
+7,1479
+4,1536
+6,1598
+6,1635
+11,1667
+4,1696
+7,1792
+7,1795
+4,1808
+6,1838
+16,1838
+3,1847
+8,1933
+8,1959
+4,1990
+9,2003
+10,2070
+7,2091
+8,2099
+5,2104
+11,2147
+4,2154
+12,2163
+11,2172
+9,2174
+13,2183
+17,2193
+11,2210
+10,2212
+4,2236
+4,2245
+8,2261
+6,2317
+8,2333
+16,2393
+10,2404
+4,2419
+11,2462
+10,2476
+11,2477
+9,2483
+11,2511
+14,2591
+6,2624
+8,2690
+12,2731
+15,2735
+9,2736
+13,2747
+18,2782
+15,2783
+12,2793
+11,2891
+12,2894
+12,2906
+14,2929
+12,2935
+3,2962
+5,3054
+7,3112
+9,3118
+11,3185
+14,3217
+18,3236
+11,3290
+11,3314
+4,3316
+13,3401
+10,3409
+10,3426
+9,3470
+11,3488
+12,3511
+4,3549
+16,3571
+20,3578
+5,3620
+15,3654
+15,3680
+12,3683
+7,3688
+7,3706
+12,3733
+21,3800
+16,3802
+13,3832
+16,3863
+8,3891
+12,4008
+20,4093
+21,4149
+15,4162
+13,4223
+13,4232
+10,4312
+22,4329
+14,4331
+16,4399
+13,4470
+24,4618
+27,4669
+16,4681
+28,4737
+11,4784
+12,4829
+14,4857
+26,4918
+27,4967
+17,5041
+20,5051
+12,5077
+20,5107
+12,5108
+24,5124
+27,5156
+25,5167
+19,5211
+21,5246
+8,5743
+15,5773
+22,5932
+21,5983
+37,5989
+23,5998
+18,6021
+25,6035
+26,6074
+17,6134
+27,6175
+13,6220
+13,6296
+15,6445
+33,6624
+24,6841
+23,6868
+18,6903
+24,6904
+21,6916
+32,6934
+23,6978
+32,7014
+16,7025
+29,7031
+33,7115
+20,7256
+19,7288
+27,7304
+10,7367
+34,7376
+36,7407
+26,7408
+33,7503
+24,7599
+37,7743
+34,7760
+37,7910
+20,7917
+28,7957
+30,7984
+27,8004
+45,8208
+39,8249
+29,8289
+22,8313
+27,8377
+19,8396
+30,8468
+34,8493
+35,8531
+21,8773
+18,8866
+41,9091
+34,9215
+51,9225
+30,9243
+32,9435
+38,9445
+18,9468
+42,9563
+60,9605
+19,9841
+29,9994
+17,10033
+29,10049
+41,10144
+31,10303
+35,10416
+27,10461
+37,10670
+18,10844
+41,10875
+39,10890
+41,11105
+61,11622
+46,12038
+47,12173
+36,12181
+43,12608
+45,12775
+46,12915
+45,13021
+49,13142
+55,13206
+64,13407
+64,13647
+66,13870
+57,13989
+53,14089
+51,14197
+36,14620
+28,14816
+59,14952
+39,15039
+73,15049
+41,15179
+48,15204
+37,16161
+72,16239
+72,16427
+48,16462
+62,16793
+51,16925
+71,17027
+60,17201
+70,17526
+59,17666
+91,17692
+52,17742
+65,18482
+77,18731
+84,18835
+51,19274
+66,19818
+53,19906
+58,20065
+75,20140
+88,20268
+83,20539
+48,20639
+69,20969
+41,21353
+73,21757
+79,22811
+63,23245
+90,23258
+92,24296
+60,24351
+63,24692
+63,24896
+75,25275
+70,25405
+90,25715
+111,26245
+103,26408
+117,26691
+118,28024
+40,28270
+83,28477
+90,29254
+97,29422
+92,30125
+104,30538
+96,34109
+142,35112
+105,35876
+145,36307
+160,39023
+127,40756
+169,42997
+104,47672
+179,49126
+152,53464
+163,56529
+167,59634
+302,60161
+246,62398
+236,62652
+250,62931
+267,63476
+244,66676
+248,74005
+360,88456

.venv/lib/python3.12/site-packages/statsmodels/datasets/cancer/data.py

@@ -0,0 +1,49 @@
+"""Breast Cancer Data"""
+from statsmodels.datasets import utils as du
+
+__docformat__ = 'restructuredtext'
+
+COPYRIGHT   = """???"""
+TITLE       = """Breast Cancer Data"""
+SOURCE      = """
+This is the breast cancer data used in Owen's empirical likelihood.  It is taken from
+Rice, J.A. Mathematical Statistics and Data Analysis.
+http://www.cengage.com/statistics/discipline_content/dataLibrary.html
+"""
+
+DESCRSHORT  = """Breast Cancer and county population"""
+
+DESCRLONG   = """The number of breast cancer observances in various counties"""
+
+#suggested notes
+NOTE        = """::
+
+    Number of observations: 301
+    Number of variables: 2
+    Variable name definitions:
+
+        cancer - The number of breast cancer observances
+        population - The population of the county
+
+"""
+
+
+def load_pandas():
+    data = _get_data()
+    return du.process_pandas(data, endog_idx=0, exog_idx=None)
+
+
+def load():
+    """
+    Load the data and return a Dataset class instance.
+
+    Returns
+    -------
+    Dataset
+        See DATASET_PROPOSAL.txt for more information.
+    """
+    return load_pandas()
+
+
+def _get_data():
+    return du.load_csv(__file__, 'cancer.csv', convert_float=True)