Changes in Customer Retention

User/views.py

@@ -654,18 +654,28 @@ def Customer_Frequency(request):
             # Loading the data
             df = pd.read_csv('Data/CustomerDatasetNew_4.csv')
 
+            # Data pre processing
+
             # Drop the missing data in dataframe
             df_data = df.dropna()
 
             # Convert the date field, InvoiceDate to datetime object
             df_data.InvoiceDate = pd.to_datetime(df_data.InvoiceDate)
 
-            ctm_bhvr_dt = df_data[(df_data.InvoiceDate < pd.Timestamp(2021, 12, 1))  # dataset End date
-                                  & (df_data.InvoiceDate >= pd.Timestamp(2021, 5, 1))].reset_index(
+            # Get least recent date
+            least_recent_date = pd.Timestamp((df_data.InvoiceDate).iloc[0])
+            updated_least_recent_date = least_recent_date.strftime('%Y,%d,%m')
+
+            # Get most recent date
+            most_recent_date = pd.Timestamp((df_data.InvoiceDate).iloc[-1])
+            updated_most_recent_date = most_recent_date.strftime('%Y,%d,%m')
+
+            ctm_bhvr_dt = df_data[(df_data.InvoiceDate < pd.Timestamp(updated_most_recent_date))  # dataset End date
+                                  & (df_data.InvoiceDate >= pd.Timestamp(updated_least_recent_date))].reset_index(
                 drop=True)  # dataset start date
 
-            ctm_next_quarter = df_data[(df_data.InvoiceDate < pd.Timestamp(2022, 3, 1))  # predict date Till
-                                       & (df_data.InvoiceDate >= pd.Timestamp(2021, 12, 1))].reset_index(
+            ctm_next_quarter = df_data[(df_data.InvoiceDate < pd.Timestamp(2022, 12, 1))  # predict date Till
+                                       & (df_data.InvoiceDate >= pd.Timestamp(updated_most_recent_date))].reset_index(
                 drop=True)  # dataset End date
 
             # Get the distinct customers in the dataframe ctm_bhvr_dt
@@ -674,44 +684,47 @@ def Customer_Frequency(request):
             # Rename the column to CustomerID.
             ctm_dt.columns = ['CustomerID']
 
-            # Create a dataframe with CustomerID and customers first purchase
-            ctm_1st_purchase_in_next_quarter = ctm_next_quarter.groupby('CustomerID').InvoiceDate.min().reset_index()
-            ctm_1st_purchase_in_next_quarter.columns = ['CustomerID', 'MinPurchaseDate']
+            # Create a dataframe with CustomerID and customers first service
+            ctm_1st_service_in_next_quarter = ctm_next_quarter.groupby('CustomerID').InvoiceDate.min().reset_index()
+            ctm_1st_service_in_next_quarter.columns = ['CustomerID', 'MinServiceDate']
 
-            # Create a dataframe with CustomerID and customers last purchase
-            ctm_last_purchase_bhvr_dt = ctm_bhvr_dt.groupby('CustomerID').InvoiceDate.max().reset_index()
-            ctm_last_purchase_bhvr_dt.columns = ['CustomerID', 'MaxPurchaseDate']
+            # Create a dataframe with CustomerID and customers last service
+            ctm_last_service_bhvr_dt = ctm_bhvr_dt.groupby('CustomerID').InvoiceDate.max().reset_index()
+            ctm_last_service_bhvr_dt.columns = ['CustomerID', 'MaxServiceDate']
 
             # Merge two dataframes
-            ctm_purchase_dates = pd.merge(ctm_last_purchase_bhvr_dt, ctm_1st_purchase_in_next_quarter, on='CustomerID',
+            ctm_service_dates = pd.merge(ctm_last_service_bhvr_dt, ctm_1st_service_in_next_quarter, on='CustomerID',
                                           how='left')
 
-            # Get the difference in days from MinPurchaseDate and MaxPurchaseDate for each customer
-            ctm_purchase_dates['NextPurchaseDay'] = (
-                    ctm_purchase_dates['MinPurchaseDate'] - ctm_purchase_dates['MaxPurchaseDate']).dt.days
+            # Get the difference in days from MinServiceDate and MaxServiceDate for each customer
+            ctm_service_dates['NextServiceDay'] = (
+                    ctm_service_dates['MinServiceDate'] - ctm_service_dates['MaxServiceDate']).dt.days
 
-            # Update the dataframe ctm_dt by merging it with the NextPurchaseDay column of the dataframe ctm_purchase_dates
-            ctm_dt = pd.merge(ctm_dt, ctm_purchase_dates[['CustomerID', 'NextPurchaseDay']], on='CustomerID',
+            # Update the dataframe ctm_dt by merging it with the NextServiceDay column of the dataframe ctm_service_dates
+            ctm_dt = pd.merge(ctm_dt, ctm_service_dates[['CustomerID', 'NextServiceDay']], on='CustomerID',
                               how='left')
-            # Fill all missing values in the dataset ctm_dt with the number 9999
+
+            # Fill all missing values in the dataset ctm_dt with 'Will not return' value
             ctm_dt = ctm_dt.fillna('Will not return')
 
             # Feature Engineering
+            # RFM Model
+
             # Recency
-            # Get the maximum purchase date of each customer and create a dataframe with it together with the customer's id.
-            ctm_max_purchase = ctm_bhvr_dt.groupby('CustomerID').InvoiceDate.max().reset_index()
-            ctm_max_purchase.columns = ['CustomerID', 'MaxPurchaseDate']
+            # Get the maximum service date of each customer and create a dataframe with it together with the customer's id.
+            ctm_max_service = ctm_bhvr_dt.groupby('CustomerID').InvoiceDate.max().reset_index()
+            ctm_max_service.columns = ['CustomerID', 'MaxServiceDate']
 
             # Find the recency of each customer in days
-            ctm_max_purchase['Recency'] = (
-                    ctm_max_purchase['MaxPurchaseDate'].max() - ctm_max_purchase['MaxPurchaseDate']).dt.days
+            ctm_max_service['Recency'] = (
+                    ctm_max_service['MaxServiceDate'].max() - ctm_max_service['MaxServiceDate']).dt.days
 
             # Find the recency in days
-            ctm_max_purchase['Recency'] = (
-                    ctm_max_purchase['MaxPurchaseDate'].max() - ctm_max_purchase['MaxPurchaseDate']).dt.days
+            ctm_max_service['Recency'] = (
+                    ctm_max_service['MaxServiceDate'].max() - ctm_max_service['MaxServiceDate']).dt.days
 
-            # Merge the dataframes ctm_dt and ctm_max_purchase[['CustomerID', 'Recency']] on the CustomerID column.
-            ctm_dt = pd.merge(ctm_dt, ctm_max_purchase[['CustomerID', 'Recency']], on='CustomerID')
+            # Merge the dataframes ctm_dt and ctm_max_service[['CustomerID', 'Recency']] on the CustomerID column.
+            ctm_dt = pd.merge(ctm_dt, ctm_max_service[['CustomerID', 'Recency']], on='CustomerID')
             Recency = ctm_dt.head()
 
             # Frequency
@@ -722,7 +735,7 @@ def Customer_Frequency(request):
             # Update the dataframe ctm_dt by merging it with the dataframe ctm_frequency
             ctm_dt = pd.merge(ctm_dt, ctm_frequency, on='CustomerID')
             Frequency = ctm_dt.head()
-            print(Frequency)
+            #print(Frequency)
             df_data['Revenue'] = df_data.ServicePrice
 
             # df_data['Revenue'] = df_data.UnitPrice * df_data.Quantity
@@ -736,7 +749,7 @@ def Customer_Frequency(request):
             pntge = (singlecustoemrvalue / totalALlcustomervalue) * 100
             sss = pd.DataFrame(pntge, columns=['persontageImpact'])
             xxx = sss.head()
-            print(xxx)
+            #print(xxx)
             ctm_dt = pd.merge(ctm_dt, sss, on='CustomerID')
             Frequency = ctm_dt.head()
             df2 = pd.read_csv('Data/CustomerDatasetNew_4.csv')
@@ -746,7 +759,7 @@ def Customer_Frequency(request):
             df0 = df0.groupby('CustomerID')[
                 'ServicePrice', 'ServiceQualityRating', 'ResponseTimeRating', 'CustomerServiceRating'].sum()
             df0 = pd.merge(df0, Frequency, on='CustomerID')
-            print(df0)
+            #print(df0)
             x = df0[['ServicePrice', 'ServiceQualityRating', 'ResponseTimeRating', 'CustomerServiceRating']]
             y = df0[['Frequency']]
             # split dataset for train and test
@@ -760,7 +773,7 @@ def Customer_Frequency(request):
             customerid = myForm.cleaned_data['customerid']
             dfz = ctm_dt[(ctm_dt["CustomerID"] == customerid)]
             Frequency = dfz.head(1)
-            print(Frequency)
+            #print(Frequency)
             df = Frequency
             json_records = df.reset_index().to_json(orient='records')
             arr = json.loads(json_records)

templates/Admin/Customer_Frequency.html

@@ -86,7 +86,7 @@
                                 <tr>
                                     <th>Customer ID  </th>
                                     <th>Customer Name  </th>
-                                    <th>Next Purchase In </th>
+                                    <th>Next Service In </th>
                                     <th>Purchased Frequency </th>
                                     <th>Impact on Profitability </th>
                                 </tr>
@@ -97,7 +97,7 @@
                                 <tr>
                                     <td>{{i.CustomerID}} </td>
                                     <td>{{i.CustomerName}} </td>
-                                    <td>{{i.NextPurchaseDay}} </td>
+                                    <td>{{i.NextServiceDay}} </td>
                                     <td>{{i.Frequency}} </td>
                                     <td>{{i.persontageImpact}} % </td>
                                 </tr>