Spaces:
Sleeping
Sleeping
add chart with proba
Browse files
app.py
CHANGED
@@ -2,6 +2,7 @@ import gradio as gr
|
|
2 |
import joblib
|
3 |
import spacy
|
4 |
import numpy as np
|
|
|
5 |
from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer
|
6 |
from sklearn.preprocessing import MultiLabelBinarizer
|
7 |
from sklearn.base import BaseEstimator, TransformerMixin
|
@@ -21,31 +22,84 @@ def lemmatize(s: str) -> iter:
|
|
21 |
# lemmatize
|
22 |
return map(lambda token: token.lemma_.lower(), tokens)
|
23 |
|
24 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
text = title + " " + post
|
26 |
lemmes = np.array([' '.join(list(lemmatize(text)))])
|
27 |
|
28 |
X = tfidf.transform(lemmes)
|
29 |
|
30 |
-
|
31 |
-
|
32 |
-
tags = list(dict(sorted(tags_binarizer.ts.count.items())).keys())
|
33 |
-
|
34 |
-
result = list(zip(tags, y_proba))
|
35 |
-
else:
|
36 |
-
y_bin = model.predict(X)
|
37 |
-
y_tags = tags_binarizer.inverse_transform(y_bin)
|
38 |
|
39 |
-
|
|
|
40 |
|
41 |
-
return
|
42 |
|
43 |
demo = gr.Interface(
|
44 |
fn=predict,
|
45 |
inputs=[
|
46 |
gr.Textbox(label="Title", lines=1, placeholder="Title..."),
|
47 |
-
gr.Textbox(label="Post", lines=
|
48 |
-
|
49 |
-
outputs=gr.Textbox(lines=10))
|
50 |
|
51 |
demo.launch()
|
|
|
2 |
import joblib
|
3 |
import spacy
|
4 |
import numpy as np
|
5 |
+
import matplotlib.pyplot as plt
|
6 |
from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer
|
7 |
from sklearn.preprocessing import MultiLabelBinarizer
|
8 |
from sklearn.base import BaseEstimator, TransformerMixin
|
|
|
22 |
# lemmatize
|
23 |
return map(lambda token: token.lemma_.lower(), tokens)
|
24 |
|
25 |
+
def plot(tags, proba):
|
26 |
+
plt.style.use('dark_background')
|
27 |
+
plt.rcParams.update({'font.size': 16})
|
28 |
+
|
29 |
+
fig, ax = plt.subplots(figsize=(12,9))
|
30 |
+
|
31 |
+
ax.barh(tags, proba, align='center', color='darkred')
|
32 |
+
ax.set_yticks(tags, labels=tags)
|
33 |
+
ax.invert_yaxis() # labels read top-to-bottom
|
34 |
+
ax.set_xlabel('Score')
|
35 |
+
ax.set_title('Score/Tag')
|
36 |
+
|
37 |
+
for i, v in enumerate(proba):
|
38 |
+
ax.text(v - 0.065, i + 0.05, str(round(v, 2)))
|
39 |
+
|
40 |
+
plt.xlim(0, 1)
|
41 |
+
plt.show()
|
42 |
+
|
43 |
+
def predict_words(X):
|
44 |
+
y_bin = model.predict(X)
|
45 |
+
y_tags = " ".join(tags_binarizer.inverse_transform(y_bin)[0])
|
46 |
+
|
47 |
+
return y_tags
|
48 |
+
|
49 |
+
def proba_chart(X):
|
50 |
+
y_proba = model.predict_proba(X)[0]
|
51 |
+
tags = list(dict(sorted(tags_binarizer.ts.count.items())).keys())
|
52 |
+
|
53 |
+
# combine
|
54 |
+
data = list(zip(tags, y_proba))
|
55 |
+
|
56 |
+
# sort
|
57 |
+
data = sorted(data, key=lambda tag_value: tag_value[1], reverse=True)
|
58 |
+
|
59 |
+
# keep values >= min_score
|
60 |
+
data = list(filter(lambda tag_value: tag_value[1] >= 0.1, data))
|
61 |
+
|
62 |
+
# we have our two dimensions for chart
|
63 |
+
tags, proba = zip(*data)
|
64 |
+
|
65 |
+
# build chart
|
66 |
+
plt.style.use('dark_background')
|
67 |
+
plt.rcParams.update({'font.size': 16})
|
68 |
+
|
69 |
+
fig, ax = plt.subplots(figsize=(12,9))
|
70 |
+
|
71 |
+
ax.barh(tags, proba, align='center', color='darkred')
|
72 |
+
ax.set_yticks(tags, labels=tags)
|
73 |
+
ax.invert_yaxis() # labels read top-to-bottom
|
74 |
+
ax.set_xlabel('Score')
|
75 |
+
ax.set_title('Score/Tag')
|
76 |
+
|
77 |
+
for i, v in enumerate(proba):
|
78 |
+
ax.text(v - 0.065, i + 0.05, str(round(v, 2)))
|
79 |
+
|
80 |
+
plt.xlim(0, 1)
|
81 |
+
|
82 |
+
return fig
|
83 |
+
|
84 |
+
def predict(title: str , post: str):
|
85 |
text = title + " " + post
|
86 |
lemmes = np.array([' '.join(list(lemmatize(text)))])
|
87 |
|
88 |
X = tfidf.transform(lemmes)
|
89 |
|
90 |
+
# predicted words
|
91 |
+
words = predict_words(X)
|
|
|
|
|
|
|
|
|
|
|
|
|
92 |
|
93 |
+
# proba chart
|
94 |
+
chart = proba_chart(X)
|
95 |
|
96 |
+
return words, chart
|
97 |
|
98 |
demo = gr.Interface(
|
99 |
fn=predict,
|
100 |
inputs=[
|
101 |
gr.Textbox(label="Title", lines=1, placeholder="Title..."),
|
102 |
+
gr.Textbox(label="Post", lines=20, placeholder="Post...")],
|
103 |
+
outputs=[gr.Textbox(label="Tags"), gr.Plot()])
|
|
|
104 |
|
105 |
demo.launch()
|