Valeriy Sinyukov commited on
Commit
bd0a951
·
1 Parent(s): ce32a21

Don't show categories with less then 1% confidence

Browse files
Files changed (1) hide show
  1. results.py +7 -6
results.py CHANGED
@@ -4,19 +4,20 @@ import pandas as pd
4
  from languages import en, ru
5
 
6
 
7
- def filter_results(results : pd.DataFrame, top_k=0.95):
 
 
 
 
8
  certs = results[certainty_init].to_numpy()
9
  cum_certs = certs.cumsum()
10
  do_no_show = cum_certs < top_k
11
  first_not_to_show_id = np.argmin(do_no_show) + 1
12
  results = results.iloc[:first_not_to_show_id]
 
13
  return results
14
 
15
 
16
- category_init = "label"
17
- certainty_init = "score"
18
-
19
-
20
  def process_keys(results: pd.DataFrame, lang):
21
  category = {en: "Category", ru: "Категория"}
22
  certainty = {en: "Certainty", ru: "Уверенность"}
@@ -121,7 +122,7 @@ def process_results(results, lang):
121
  results = pd.DataFrame(results)
122
  results = process_categories(results, lang)
123
  results = results.groupby(by=category_init, as_index=False).sum()
124
- results = results.sort_values(by=[certainty_init],ascending=False)
125
  results = filter_results(results)
126
  results = process_certainities(results)
127
  results = process_keys(results, lang)
 
4
  from languages import en, ru
5
 
6
 
7
+ category_init = "label"
8
+ certainty_init = "score"
9
+
10
+
11
+ def filter_results(results: pd.DataFrame, top_k=0.95):
12
  certs = results[certainty_init].to_numpy()
13
  cum_certs = certs.cumsum()
14
  do_no_show = cum_certs < top_k
15
  first_not_to_show_id = np.argmin(do_no_show) + 1
16
  results = results.iloc[:first_not_to_show_id]
17
+ results = results[results[certainty_init] >= 0.01]
18
  return results
19
 
20
 
 
 
 
 
21
  def process_keys(results: pd.DataFrame, lang):
22
  category = {en: "Category", ru: "Категория"}
23
  certainty = {en: "Certainty", ru: "Уверенность"}
 
122
  results = pd.DataFrame(results)
123
  results = process_categories(results, lang)
124
  results = results.groupby(by=category_init, as_index=False).sum()
125
+ results = results.sort_values(by=[certainty_init], ascending=False)
126
  results = filter_results(results)
127
  results = process_certainities(results)
128
  results = process_keys(results, lang)