if options.has_class:
evals_dict0 = {}
evals_dict0['Grant.Status'] = [int(s) for s in evals_dict['Grant.Status']]
if DO_COMPOUND_RULES:
print 'compound_rules:', len(compound_rules)
print 'num_rules:', num_rules
for i, compound in enumerate(compound_rules[:num_rules]):
attrs = [attr for (attr, _, _) in compound]
val_rows = [[data_dict[attr][instance] for attr in attrs] for instance in range(num_instances)]
evals = [1 if evaluate_compound_rule(compound, vals) else 0 for vals in val_rows]
evals_dict0[compound_rule_to_string(compound)] = evals
else:
for i, rule in enumerate(sorted_keys[:num_rules]):
attr, _, _ = rule
vals = data_dict[attr]
evals = [1 if evaluate_rule(rule, val) else 0 for val in vals]
evals_dict0[rule_to_string(rule)] = evals
analyse_evals_dict(evals_dict0, evals_header)
csv.writeCsvDict(knn_file_csv, evals_dict, evals_header)
if False:
out_filename = filename + '.csv'
out_lines = ['Grant.Application.ID,Grant.Status,Success']
for k in sorted(results.keys()):
r = results[k]
out_lines.append(','.join([str(x) for x in [k, r['prob1'], r['predicted1']]]))
out_data = '\n'.join(out_lines)
file(out_filename, 'wt').write(out_data)
}
if False:
histo = getHistogram(sales_mo_filtered, measured_keys, stats, 20000)
exit()
if False:
keys = [k+':'+t for k in measured_keys for t in ['level','number']]
columns = [histo[k][i] for k in measured_keys for i in [0,1]]
print len(columns), [len(c) for c in columns]
histo_cols = dict(zip(keys, columns))
if False:
histo_cols = {}
for k in measured_keys:
for i in [0,1]:
ck = k + ' ' + ['level','number'][i]
histo_cols[ck] = [histo[k][n][i] for n in range(len(histo[k]))]
csv.writeCsvDict(sales_mo_histo, histo_cols)
if False:
print zip(h1, h2)
print [x1 == x2 for x1,x2 in zip(h1,h2)]
print '======================================'
if False:
groups = {}
for h in h1:
parts = h.split('_')
if len(parts) >= 2:
groups[parts[0]] = []
else:
groups[h] = [h]
for h in h1:
