added touchpoint

.ipynb_checkpoints/app-checkpoint.py

@@ -0,0 +1,58 @@
+import streamlit as st
+import plotly.express as px
+from plotly.subplots import make_subplots
+from utils import *
+
+########## Title for the Web App ##########
+st.title("Text Classification for HC")
+
+########## Create Input field ##########
+feedback = st.text_input('Type your text here', 'Customer suggested that the customer service needs to be improved and the response time needs to be improved.')
+
+if st.button('Click for predictions!'):
+    with st.spinner('Generating predictions...'):
+        
+        topics_prob, sentiment_prob, touchpoint_prob = get_single_prediction(feedback)
+        
+        bar_topic = px.bar(topics_prob, x='probability', y='topic')
+        
+        bar_touchpoint = px.bar(touchpoint_prob, x='probability', y='touchpoint')
+                
+        pie = px.pie(sentiment_prob, 
+               values='probability', 
+               names='sentiment', 
+               color_discrete_map={'positive':'rgb(0, 204, 0)', 
+                                 'negative':'rgb(215, 11, 11)'
+                                  },
+               color='sentiment'
+              )
+        
+    st.plotly_chart(bar, use_container_width=True)    
+    st.plotly_chart(pie, use_container_width=True)
+
+st.write("\n")    
+st.subheader('Or... Upload a csv file if you have a file instead.')
+st.write("\n")
+
+st.download_button(
+     label="Download sample file here",
+     data=sample_file,
+     file_name='sample_data.csv',
+     mime='text/csv',
+ )
+
+uploaded_file = st.file_uploader("Please upload a csv file with only 1 column of texts.")
+
+if uploaded_file is not None:
+    
+    with st.spinner('Generating predictions...'):
+        results = get_multiple_predictions(uploaded_file)
+    
+    st.download_button(
+     label="Download results as CSV",
+     data=results,
+     file_name='results.csv',
+     mime='text/csv',
+     )
+    
+    

.ipynb_checkpoints/utils-checkpoint.py

@@ -3,11 +3,12 @@ import pickle
 import numpy as np
 import pandas as pd
 
-
+svc = pickle.load(open('models/svc_model.sav', 'rb'))
 tfidf = pickle.load(open('models/tfidf.sav', 'rb'))
 svc_sentiment = pickle.load(open('models/sentiment_model.sav', 'rb'))
 tfidf_sentiment = pickle.load(open('models/tfidf_sentiment.sav', 'rb'))
-svc = pickle.load(open('models/svc_model.sav', 'rb'))
+svc_touchpoint = pickle.load(open('models/touchpoint_model.sav', 'rb'))
+tfidf_touchpoint = pickle.load(open('models/tfidf_touchpoint.sav', 'rb'))
 
 labels = [
        'Product quality', 'Knowledge',
@@ -23,23 +24,24 @@ def get_single_prediction(text):
     
     # manipulate data into a format that we pass to our model
     text = text.lower().strip() #lower case
-
-    # Vectorise text
-    text_vectors = tfidf.transform([text])
-
+    
     # Make topic predictions
+    text_vectors = tfidf.transform([text])
     results = svc.predict_proba(text_vectors).squeeze().round(2)
-
     pred_prob = pd.DataFrame({'topic': labels, 'probability': results}).sort_values('probability', ascending=True)
 
     # Make sentiment predictions
     text_vectors_sentiment = tfidf_sentiment.transform([text])
  
-    results_sentiment = svc_sentiment.predict_proba(text_vectors).squeeze().round(2)
+    results_sentiment = svc_sentiment.predict_proba(text_vectors_sentiment).squeeze().round(2)
     pred_prob_sentiment = pd.DataFrame({'sentiment': ['Negative', 'Positive'], 'probability': results_sentiment}).sort_values('probability', ascending=True)
     
-
-    return (pred_prob, pred_prob_sentiment)
+    # Make touchpoint predictions
+    text_vectors_touchpoint = tfidf_touchpoint.transform([text])
+    results_touchpoint = svc_touchpoint.predict_proba(text_vectors_touchpoint).squeeze().round(2)
+    pred_prob_touchpoint = pd.DataFrame({'touchpoint': ['ASC', 'CC', 'Technician'], 'probability': results_touchpoint}).sort_values('probability', ascending=True)
+    
+    return (pred_prob, pred_prob_sentiment, pred_prob_touchpoint)
 
 def get_multiple_predictions(csv):
     
@@ -65,9 +67,13 @@ def get_multiple_predictions(csv):
     # Vectorise text and get sentiment predictions 
     text_vectors_sentiment = tfidf_sentiment.transform(df['sequence_clean'])
     pred_results_sentiment = pd.DataFrame(svc_sentiment.predict(text_vectors_sentiment), columns = ['sentiment'])    
+
+    # Vectorise text and get touchpoint predictions 
+    text_vectors_touchpoint = tfidf_touchpoint.transform(df['sequence_clean'])
+    pred_results_touchpoint = pd.DataFrame(svc_touchpoint.predict(text_vectors_touchpoint), columns = ['touchpoint'])
     
     # Join back to original sequence
-    final_results = df.join(pred_results).join(pred_results_sentiment)
+    final_results = df.join(pred_results).join(pred_results_sentiment).join(pred_results_touchpoint)
     
     final_results.drop(columns=['sequence_clean'], inplace=True)
     

app.py

@@ -12,9 +12,11 @@ feedback = st.text_input('Type your text here', 'Customer suggested that the cus
 if st.button('Click for predictions!'):
     with st.spinner('Generating predictions...'):
         
-        topics_prob, sentiment_prob = get_single_prediction(feedback)
+        topics_prob, sentiment_prob, touchpoint_prob = get_single_prediction(feedback)
         
-        bar = px.bar(topics_prob, x='probability', y='topic')
+        bar_topic = px.bar(topics_prob, x='probability', y='topic')
+        
+        bar_touchpoint = px.bar(touchpoint_prob, x='probability', y='touchpoint')
                 
         pie = px.pie(sentiment_prob, 
                values='probability', 

utils.py

@@ -3,11 +3,12 @@ import pickle
 import numpy as np
 import pandas as pd
 
-
+svc = pickle.load(open('models/svc_model.sav', 'rb'))
 tfidf = pickle.load(open('models/tfidf.sav', 'rb'))
 svc_sentiment = pickle.load(open('models/sentiment_model.sav', 'rb'))
 tfidf_sentiment = pickle.load(open('models/tfidf_sentiment.sav', 'rb'))
-svc = pickle.load(open('models/svc_model.sav', 'rb'))
+svc_touchpoint = pickle.load(open('models/touchpoint_model.sav', 'rb'))
+tfidf_touchpoint = pickle.load(open('models/tfidf_touchpoint.sav', 'rb'))
 
 labels = [
        'Product quality', 'Knowledge',
@@ -23,23 +24,24 @@ def get_single_prediction(text):
     
     # manipulate data into a format that we pass to our model
     text = text.lower().strip() #lower case
-
-    # Vectorise text
-    text_vectors = tfidf.transform([text])
-
+    
     # Make topic predictions
+    text_vectors = tfidf.transform([text])
     results = svc.predict_proba(text_vectors).squeeze().round(2)
-
     pred_prob = pd.DataFrame({'topic': labels, 'probability': results}).sort_values('probability', ascending=True)
 
     # Make sentiment predictions
     text_vectors_sentiment = tfidf_sentiment.transform([text])
  
-    results_sentiment = svc_sentiment.predict_proba(text_vectors).squeeze().round(2)
+    results_sentiment = svc_sentiment.predict_proba(text_vectors_sentiment).squeeze().round(2)
     pred_prob_sentiment = pd.DataFrame({'sentiment': ['Negative', 'Positive'], 'probability': results_sentiment}).sort_values('probability', ascending=True)
     
-
-    return (pred_prob, pred_prob_sentiment)
+    # Make touchpoint predictions
+    text_vectors_touchpoint = tfidf_touchpoint.transform([text])
+    results_touchpoint = svc_touchpoint.predict_proba(text_vectors_touchpoint).squeeze().round(2)
+    pred_prob_touchpoint = pd.DataFrame({'touchpoint': ['ASC', 'CC', 'Technician'], 'probability': results_touchpoint}).sort_values('probability', ascending=True)
+    
+    return (pred_prob, pred_prob_sentiment, pred_prob_touchpoint)
 
 def get_multiple_predictions(csv):
     
@@ -65,9 +67,13 @@ def get_multiple_predictions(csv):
     # Vectorise text and get sentiment predictions 
     text_vectors_sentiment = tfidf_sentiment.transform(df['sequence_clean'])
     pred_results_sentiment = pd.DataFrame(svc_sentiment.predict(text_vectors_sentiment), columns = ['sentiment'])    
+
+    # Vectorise text and get touchpoint predictions 
+    text_vectors_touchpoint = tfidf_touchpoint.transform(df['sequence_clean'])
+    pred_results_touchpoint = pd.DataFrame(svc_touchpoint.predict(text_vectors_touchpoint), columns = ['touchpoint'])
     
     # Join back to original sequence
-    final_results = df.join(pred_results).join(pred_results_sentiment)
+    final_results = df.join(pred_results).join(pred_results_sentiment).join(pred_results_touchpoint)
     
     final_results.drop(columns=['sequence_clean'], inplace=True)