Upload 12 files

app.py

@@ -17,13 +17,11 @@ from PIL import Image
 
 # Create sidebar navigation
 
-## Churn Guardian, Yosef Project
-
 # st.markdown(
 #     f"""
 #     <style>
 #         [data-testid="stSidebar"] {{
-#             background-image: url(https://raw.githubusercontent.com/FTDS-assignment-bay/FTDS-007-HCK-group-002/main/assets/ChurnGuardian-Logo-Transparants.png);
+#             background-image: url(https://raw.githubusercontent.com/FTDS-assignment-bay/main/assets/ChurnGuardian-Logo-Transparants.png);
 #             background-repeat: no-repeat;
 #             padding-top: 20px;
 #             background-position: 10px 50px;
@@ -34,9 +32,6 @@ from PIL import Image
 #     unsafe_allow_html=True,
 # )
 
-# selected_page = st.sidebar.radio('Select Page', ('📋 Home Page', '📊 Exploratory Data Analysis', '💻 Model'))
-## Churn Guardian, Yosef Project
-
 st.set_page_config(
     page_title='Telco Customer Churn and Segmentation', 
     layout='centered', #wide
@@ -57,7 +52,7 @@ selected = option_menu(None, ["About", "EDA", "Predict"],
     menu_icon="cast", default_index=0, orientation="horizontal",
     styles={
         "icon": {"color": "orange", "font-size": "15px"}, 
-        "nav-link": {"font-size": "15px", "text-align": "left", "margin":"1px", "--hover-color": "#eee"},
+        "nav-link": {"font-size": "15px", "text-align": "left", "margin":"1px", "--hover-color": "#eee"}, 
         "nav-link-selected": {"background-color": "grey"},
     }
 )   

eda.py

@@ -6,7 +6,7 @@ import plotly.express as px
 from PIL import Image
 
 def run():
-    #Show dataframe
+    # Show dataframe
     # st.title('Data Overview')
     df = pd.read_csv('telco_data_clean.csv')
     # st.dataframe(df.head())
@@ -14,76 +14,77 @@ def run():
     st.title('Exploratory Data Analysis')
     plot_selection = st.selectbox(label='Choose', 
                                   options=['Customer Distribution', 
-                                           'Top Total Churn City',
-                                           'Customers reasons for churning', 
-                                           'Churn Reason', 
-                                           'Age Distribution Churn vs Stayed',  
-                                           'Gender Distribution Churn vs Stayed'])
+                                           'Churn by Monthly Charge',
+                                           'Churn by Tenure', 
+                                           'Churn by Internet Service'])
     
     # Plot 1
     def plot_1():
         st.write('#### Pie Chart for Customer Status Distribution')
-        # fig_1 = plt.figure()
-        # customer_status_count = df['Customer Status'].value_counts()
-        # fig_1, ax = plt.subplots()
-        # ax.pie(customer_status_count, labels=customer_status_count.index, autopct='%1.1f%%')
-        # ax.set_title('Customer Status Distribution')
-        # st.pyplot(fig_1)
-        # with st.expander('Explanation'):
-        #     st.text('''
-        #         The data frame indicates that 26.5% of customers have churned. 
-        #         The "Joined" Category row will be removed, as it doesn't offer 
-        #         any useful insights into the churn rate.
-        #     ''')
-        st.text('''
-                The data frame indicates that 26.5% of customers have churned. 
-                The "Joined" Category row will be removed, as it doesn't offer 
-                any useful insights into the churn rate.
-
-                Sisa ne nyusul
-            ''')
-
-    # st.write('## Histogram Limit Balance')
-    # fig = plt.figure(figsize=(15,5))
-    # sns.histplot(df['limit_balance'], bins=20, kde=True).set(title='limit_balance')
-    # st.pyplot(fig)
-    # st.write('Based on histogram, column _limit\_balance_ skewness is positive, meaning the data distribution is not normal.')
-    # st.markdown('---')
-
-    # st.write('## Average Amount of Bill Statement')
-    # df_amt = pd.DataFrame()
-    # bill_amt = []
-    # pay_amt = []
-    # for i in range(1, 7):
-    #     bill_amt.append(df['bill_amt_' + str(i)].mean())
-    #     pay_amt.append(df['pay_amt_' + str(i)].mean())
-    # df_amt['bill_amt'] = bill_amt
-    # df_amt['pay_amt'] = pay_amt
-    # fig, ax = plt.subplots(ncols=2, figsize=(10, 5))
-    # axis_label = sns.barplot(ax=ax[0], data=df_amt, x=df_amt['bill_amt'].index, y=bill_amt, orient='v')
-    # ax[0].set_title('Bill Statement')
-    # axis_label = sns.barplot(ax=ax[1], data=df_amt, x=df_amt['pay_amt'].index, y=pay_amt, orient='v')
-    # ax[1].set_title('Payment')
-    # st.pyplot(fig)
-    # st.write('Average amount of bill statement is decrease every month, showing people using their credit card less during this period.')
-    # st.markdown('---')
+        target = df["churn"].value_counts().reset_index()
+        persen = df["churn"].value_counts(normalize=True).reset_index()
+        target["percentage"] = persen["churn"]
+        
+        fig_1 = plt.figure()
+        fig_1, ax = plt.subplots(ncols=1, figsize=(5, 5))
+        ax.pie(target["percentage"], labels = target["index"], autopct='%.0f%%')
+        ax.set_title("Customer Status Distribution")
+        st.pyplot(fig_1)
+        st.write('''
+                From the plot above, it is found that of the total number of customers who churn 
+                is 27% (1869 customers) and customers who is not churn / stay is 73% (5163 customers).
+                ''')
+        st.markdown('---')
+    
+    # Plot 2
+    def plot_2():
+        df_churn_by_mcharges = df.groupby(['monthly_charges_cat']).agg(total=('monthly_charges_cat', 'count')).sort_values(by=['total'], ascending=True)
+        fig_2 = plt.figure(figsize=(7, 5))
+        ax = sns.barplot(data=df_churn_by_mcharges, x=df_churn_by_mcharges.index.to_list(), y='total', orient='v')
+        ax.bar_label(ax.containers[0])
+        ax.set(title='Churn by Monthly Charges')
+        st.pyplot(fig_2)
+        st.write('''
+                 From the bar plot above we can see that Medium-High Expenses and High Expense 
+                 have the highest churn rate.
+                 ''')
+        st.markdown('---')
 
-    # st.write('## Barplot Sex')
-    # fig = plt.figure(figsize=(10,5))
-    # sns.countplot(x='sex', data=df)
-    # st.pyplot(fig)
-    # st.write('Most of the bank customer is female')
-    # st.markdown('---')
+    # Plot 3
+    def plot_3():
+        df_churn_by_mcharge = df.groupby(['tenure_year', 'churn']).agg(total=('churn', 'count'))
+        fig_3 = plt.figure(figsize=(7, 5))
+        ax = sns.lineplot(data=df_churn_by_mcharge, x="tenure_year", y="total", hue="churn")
+        # ax.bar_label(ax.containers[0])
+        ax.set(title='Churn by Monthly Charges')
+        st.pyplot(fig_3)
+        st.write('''
+                 From the bar plot above we can see that as tenure increases, the churn rate tends 
+                 to decrease. Customers with a longer usage period tend to be more loyal.
+                 ''')
+        st.markdown('---')
+    
+    # Plot 4
+    def plot_4():
+        fig_4 = plt.figure(figsize=(7, 5))
+        ax = sns.countplot(data=df, x='internet_service', hue='churn')
+        ax.bar_label(ax.containers[0])
+        ax.bar_label(ax.containers[1])
+        ax.set(title='Churn by Internet Service')
+        st.pyplot(fig_4)
+        st.write('There are more customers with Fiber Optic services who churn (1297) than DSL customers who churn (459). The ratio of churn to total customers appears to be higher for customers with Fiber Optic services (41.9%) compared to DSL customers (19.0%).')
+        st.write('The solution that needs to be done is to evaluate and update Fiber Optic services to improve quality and customer satisfaction, such as focusing on improving speed, stability and ease of use. In addition, there needs to be an adjustment to the marketing strategy to emphasize the advantages and benefits of Fiber Optic services that can meet customer needs by identifying market segments that are more likely to be interested in this service.')
+        st.markdown('---')
 
-    # st.write('## Barplot Marital Status')
-    # fig = plt.figure(figsize=(15,5))
-    # sns.countplot(x='marital_status', data=df)
-    # st.pyplot(fig)
-    # st.write('Most of the bank customer is married')
-    # st.markdown('---')
 
     if plot_selection == "Customer Distribution":
         plot_1()
+    elif plot_selection == "Churn by Monthly Charge":
+        plot_2()
+    elif plot_selection == "Churn by Tenure":
+        plot_3()
+    elif plot_selection == "Churn by Internet Service":
+        plot_4()
 
 if __name__ == '__main__':
     run()

home.py

@@ -9,7 +9,8 @@ def run():
 
     st.write('## Background :')
     st.markdown('''
-                Telco tambahono rek .....
+                The telecommunications industry, being highly competitive, faces challenges in retaining customers. 
+                Churn or customer attrition, is a critical metric that directly impacts the revenue and growth of a Telco company.
                 ''')
 
     st.write('## Project Objective :')

prediction.py

@@ -28,12 +28,14 @@ def run():
 
     # Choice of input: Upload or Manual Input
     inputType = st.selectbox("How would you like to input data ?", ["Upload Excel or CSV File", "Manual Input"])
+    st.markdown('---')
 
     # Create Function for Prediction
     def predictData(df):
         # Classification prediction
         y_pred_uploaded = classification_model.predict(df)
         df['churn'] = y_pred_uploaded
+        # st.dataframe(df)
 
         # Filter the DataFrame for Predicted Churn (1) 
         df_churn = df[df['churn'] == 1]
@@ -70,41 +72,65 @@ def run():
             df_cluster_1 = df_churn[df_churn['cluster'] == 1]
             df_cluster_2 = df_churn[df_churn['cluster'] == 2]
 
-            st.write('## Result')
-            st.write('##### Here are some suggestion to minimalize churn potential for each customer')
+            st.write(f'## Result : `{churnCustomer} customer` are predicted as churn!')
+            st.write('##### Here are some suggestion to minimalize churn potential for each customer depend on their cluster')
             c0, c1, c2 = '', '', ''
             for x in df_cluster_0['name']: c0 += str(x) + ', '
             for y in df_cluster_1['name']: c1 += str(y) + ', '
             for z in df_cluster_2['name']: c2 += str(z) + ', '
             
+            cluster_0 = '''
+                - Most of them are senior citizen
+                - Having partner and dependents
+                - High monthly charges
+            '''
+
             suggestion_0 = '''
-                - Menawarkan paket dengan tambahan kecepatan selama 3 bulan bagi yang telah berlangganan di atas 3 tahun
-                - Membuka seluruh channel TV saat event hari besar seperti lebaran, natal dan lain lain
-                - Memberikan penawaran khusus untuk meningkatkan kecepatan internet kepada mereka
+                - Offers packages with additional speed for 3 months for those who have subscribed for more than 3 years
+                - Open all TV channels during big holiday events such as Eid, Christmas and others
+                - Provide special offers to increase internet speed to them
+            '''
+
+            cluster_1 = '''
+                - Mix of senior citizan and youngster
+                - Having partner and dependents
+                - Low monthly charges
             '''
 
             suggestion_1 = '''
-                - Memberikan penawaran dengan banyak keuntungan jika berlangganan untuk jangka panjang 
-                - Menawarkan paket internet DSL tahunan dengan harga yang terjangkau
+                - Provides offers with many benefits if they subscribe for the long term
+                - Offers annual DSL internet packages at affordable prices
+            '''
+
+            cluster_2 = '''
+                - Most of them are young people
+                - Most of them have no partner and dependents
+                - Moderate monthly charges
             '''
 
             suggestion_2 = '''
-                Memberikan paket khusus dengan kriteria sebagai berikut :
-                - Kecepatan tinggi tetapi banwidth lebih rendah dengan harga yang lebih murah dari paket normal
-                - Kecepatan rendah tetapi banwidth besar sehingga koneksi jauh lebih stabil dengan harga yang lebih murah dari paket normal
+                Providing special packages with the following criteria:
+                 - High speed internet but lower bandwidth at a cheaper price than normal packages
+                 - Low speed internet but large bandwidth so the connection is much more stable at a cheaper price than normal packages
             '''
 
             if c0 != '':
+                st.write('##### Cluster 1')
+                st.write(cluster_0)
                 st.write('Suggestion for `', c0[0:-2], '` is')
                 st.write(suggestion_0)
                 st.markdown('---')
             
             if c1 != '':
+                st.write('##### Cluster 2')
+                st.write(cluster_1)
                 st.write('Suggestion for `', c1[0:-2], '` is')
                 st.write(suggestion_1)
                 st.markdown('---')
             
             if c2 != '':
+                st.write('##### Cluster 3')
+                st.write(cluster_2)
                 st.write('Suggestion for `', c2[0:-2], '` is')
                 st.write(suggestion_2)
                 st.markdown('---')
@@ -129,7 +155,15 @@ def run():
 
     # A. For CSV
     if inputType == "Upload Excel or CSV File":
-        uploaded_file = st.file_uploader("Choose a Excel or CSV file", type=["csv", "xlsx"], accept_multiple_files=False)
+        with open('telco_data_test.xlsx', 'rb') as file:
+            st.download_button(
+                label='💾 Download Template Excel',
+                data=file,
+                file_name='telco_example.xlsx',
+                mime='application/vnd.ms-excel'
+            )
+
+        uploaded_file = st.file_uploader("Choose Excel or CSV file", type=["csv", "xlsx"], accept_multiple_files=False)
         if uploaded_file is not None:
             split_file_name = os.path.splitext(uploaded_file.name)
             # file_name = split_file_name[0]
@@ -163,7 +197,7 @@ def run():
         # internet_service =  col4.selectbox('Internet Service', ('DSL', 'Fiber optic', 'No'), index=0)
 
         col4, col5, col6 = st.columns([1, 1, 1])
-        internet_service =  col4.radio(label='Subs for Phone service?', options=['DSL', 'Fiber optic', 'No'])
+        internet_service =  col4.radio(label='Subs for Internet service?', options=['DSL', 'Fiber optic', 'No'])
         phone_service = col5.radio(label='Subs for Phone service?', options=['Yes', 'No'])
         multiple_lines = col6.radio(label='Subs for Multiple Lines?', options=['Yes', 'No', 'No Phone Services'])
 
@@ -183,7 +217,7 @@ def run():
         col_charges1, col_charges2, col_charges3 = st.columns([1, 1, 2])
         monthly_charges =  col_charges1.number_input('Monthly Charges', min_value=1, max_value=999, step=1, help='Amount to paid per month', key='mcharges', on_change=calculateChargesAndCategory)
         total_charges = col_charges2.number_input('Total Charges', min_value=1, max_value=999999, step=1, disabled=True, key='tcharges')
-        charges_cat = col_charges3.text_input('Chargest Category', disabled=True, key='catcharges')
+        charges_cat = col_charges3.text_input('Charges Category', disabled=True, key='catcharges')
 
         # st.button('Predict', on_click=predict)
         data_inf = {
@@ -192,7 +226,7 @@ def run():
             'senior_citizen': senior_citizen, 
             'partner': partner, 
             'dependents': dependents, 
-            'tenure': int(tenure), 
+            'tenure': tenure, 
             'phone_service': phone_service, 
             'multiple_lines': multiple_lines, 
             'internet_service': internet_service, 
@@ -206,7 +240,7 @@ def run():
             'paperless_billing': paperless_billing, 
             'payment_method': payment_method, 
             'monthly_charges': monthly_charges, 
-            'total_charges': int(total_charges), 
+            'total_charges': total_charges, 
             'monthly_charges_cat': charges_cat, 
             'tenure_year': tenure_year
             }