feat: updated ECPM formula

requirements.txt

@@ -1,2 +1,3 @@
 seaborn
-python-dotenv
+python-dotenv
+pandas

utils/gradio_utils.py

@@ -256,9 +256,8 @@ def compute_offer(csv_file, days_lookback, min_sent, domain, offer_type, x_list)
 
     cols = ['Campanie', 'Oferta', 'Nume', 'Server', 'User',
         'Lista Custom', 'Data', 'HClicks', 'Clicks', 'Unscribers', 'Openers',
-        'Click Open', 'Leads', 'CLike', 'Complains', 'Traps', 'Send', 'ECPM']
-    # df_all = raw_df[raw_df['Nume'].str.contains('|'.join(cmp_list))]  #1
-    # df_all = df_all[df_all['Domeniu'] == 'Comcast'] #2
+        'Click Open', 'Leads', 'CLike', 'Complains', 'Traps', 'Send', 'ECPM', 'Comision', 'Domeniu']
+
     exclude_list = df_all[(df_all['Data'] > (pd.Timestamp('now') - pd.Timedelta(days=days_lookback))) \
                             & (df_all['Domeniu'] == domain)]['Oferta'].unique()
     df_all = df_all[~df_all['Oferta'].isin(exclude_list)]
@@ -270,7 +269,8 @@ def compute_offer(csv_file, days_lookback, min_sent, domain, offer_type, x_list)
     
     df_all['Click Open'] = df_all['Click Open'].str.replace('%', '').astype(float)
     df_all['ECPM'] = df_all['ECPM'].astype(float)
-
+    df_all['Comision'] = df_all['Comision'].astype(float)
+    df_all['Send'] = df_all['Send'].astype(int)
 
     # Limit the characters in the "Nume" column
     # df_all["Nume"] = df_all["Nume"].apply(_limit_chars)
@@ -318,11 +318,18 @@ def compute_offer(csv_file, days_lookback, min_sent, domain, offer_type, x_list)
         final_df.reset_index(inplace=True)
     elif offer_type == "Offers - IDs only":
         df_all['offer_id'] = df_all['Nume'].str.extract(r'(\d{3,4}$)')
+
         final_df = df_all.groupby(["offer_id"])\
-            .agg( times_sent=('offer_id', 'count'), send_avg=('Send', 'mean'), ECPM=('ECPM', 'mean'))\
-            .sort_values(['ECPM', 'times_sent'], ascending=False)
+            .agg( times_sent=('offer_id', 'count'), send_avg=('Send', 'mean'), total_sent=('Send', 'sum'),\
+                USD=('Comision', 'sum'))
+            
+        final_df['USD'] = final_df['USD'].round(2).astype(float)
         final_df['send_avg'] = final_df['send_avg'].astype(int)
+        final_df['ECPM'] = ( ( final_df['USD'] * 33.33 ) / final_df['total_sent'] ) * 1000
         final_df['ECPM'] = final_df['ECPM'].round(2).astype(float)
+
+        final_df.sort_values(by='ECPM', ascending=False, inplace=True)
+
         final_df.reset_index(inplace=True)
     else:
         final_df = pd.DataFrame()