from pubstats import h_index
import gscholar
if __name__ == '__main__':
name = sys.argv[1]
url = sys.argv[2]
year = int(sys.argv[3])
pubs,all_pubs,hidx,eidx = compute_individual_stats(name,url)
# get all citing publications
print 'Obtaining all citations'
citations = {}
for pub in pubs:
citations[pub] = gscholar.get_citations(pub)
# compute the h-index for author over productive years
now = datetime.date.today().year
X = range(year,now+1)
Y = []
for yr in range(year,now+1):
print 'YEAR=%d' % yr
for pub in pubs:
pub.cites = len(filter(lambda x: x.year is not None and x.year <= yr, citations[pub]))
Y.append(h_index(pubs))
figure()
plot(X,Y)
show()
def validate(filters=BEST_FILTERS,min_vocab_match_size=2,vocab_use_pct=1.0,replace_url_fxn=None):
"""
Run Topp on the test data using the filters and parameters specified (filters, min_vocab_match_size,vocab_use_pct).
replace_url_fxn(name,url,data) is a function that produces text content that should be used in place of the individual's URL.
The output should be a string.
"""
tmp_path = rb().data_path('validation','__TMP__REPLACE_CONTENT__.txt')
tmp_url = 'file://%s' % tmp_path
reader = csv.reader(rb().open_data(['validation','batch.txt'],'r'),delimiter=',')
senses = []
specs = []
fp_pcts = []
hindicies = []
true_hindicies = []
eindicies = []
true_eindicies = []
#print 'All_pubs Est_Pubs True_Pubs True_Pubs_1995 TP FP FN'
#print 'Name\tSense\t\tSpec\t\t%FP'
num_records = 0
for data in reader:
num_records += 1
if replace_url_fxn is None:
all_pubs,est_pubs,true_pubs,num_pubs,num_est_pubs,num_true_pubs,num_true_1995pubs,tp,fp,fn = compute_test_case_stats(*(list(data) + [filters,min_vocab_match_size,vocab_use_pct]))
else:
# get the new content
content = replace_url_fxn(*data)
# write the content to the temporary file
fh = open(tmp_path,'w')
fh.write(content)
fh.close()
all_pubs,est_pubs,true_pubs,num_pubs,num_est_pubs,num_true_pubs,num_true_1995pubs,tp,fp,fn = compute_test_case_stats(*([data[0],tmp_url,data[2]] + [filters,min_vocab_match_size,vocab_use_pct]))
hindicies.append(pubstats.h_index(est_pubs))
eindicies.append(pubstats.e_index(est_pubs))
true_hindicies.append(pubstats.h_index(true_pubs))
true_eindicies.append(pubstats.e_index(true_pubs))
tn = num_pubs - tp - fp - fn
sense = sensitivity(tp,fn)
senses.append(sense)
spec = specificity(tn,fp)
specs.append(spec)
fp_pct = float(fp) / float(tp + fp) if (tp + fp) != 0 else float('infinity')
fp_pcts.append(fp_pct)
names = data[0].split()
#print names[0][0] + ' ' + names[-1][:4],'\t','%f\t%f\t%f' % (sense,spec,fp_pct)
avg_sense = sum(senses) / float(num_records)
sdv_sense = sum([math.pow(x-avg_sense,2) for x in senses]) / float(num_records)
avg_spec = sum(specs) / float(num_records)
sdv_spec = sum([math.pow(x-avg_spec,2) for x in specs]) / float(num_records)
avg_fp_pct = sum(fp_pcts) / float(num_records)
sdv_fp_pct = sum([math.pow(x-avg_fp_pct,2) for x in fp_pcts]) / float(num_records)
print '='*80
print 'AVG\t%f\t%f\t%f' % (avg_sense,avg_spec,avg_fp_pct)
print 'SDEV\t%f\t%f\t%f' % (sdv_sense,sdv_spec,sdv_fp_pct)
return (avg_sense,sdv_sense),(avg_spec,sdv_spec),(avg_fp_pct,sdv_fp_pct),(hindicies,true_hindicies),(eindicies,true_eindicies)
