modularising data ingestion

app.py

@@ -5,23 +5,7 @@ from datasets import load_dataset
 import os
 from src.basic_plot import basic_chart
 from src.map_viz import calling_map_viz
-
-def daily_average(counts_df):
-    
-    filtered_views_list = ['View_from_Second_Link_at_Tuas_to_sg',
-                        'View_from_Second_Link_at_Tuas_to_jh',
-                        'View_from_Tuas_Checkpoint_to_sg',
-                        'View_from_Tuas_Checkpoint_to_jh',
-                        'View_from_Woodlands_Causeway_Towards_Johor_to_sg',
-                        'View_from_Woodlands_Causeway_Towards_Johor_to_jh',
-                        'View_from_Woodlands_Checkpoint_Towards_BKE_to_sg',
-                        'View_from_Woodlands_Checkpoint_Towards_BKE_to_jh']
-
-    counts_df_filter_views = counts_df[counts_df['view'].isin(filtered_views_list)]
-    counts_df_filter_views['date'] = pd.to_datetime(counts_df_filter_views['date'])
-    counts_df_filter_views['day_of_week'] = counts_df_filter_views['date'].dt.day_of_week
-    date_view_group = counts_df_filter_views.groupby(by=['view', 'day_of_week']).mean()
-    date_view_group = date_view_group.reset_index()
+from src.data_ingestion import daily_average
 
 
 def main():

src/data_ingestion.py

@@ -0,0 +1,19 @@
+import pandas as pd
+import numpy as np
+
+def daily_average(counts_df):
+    
+    filtered_views_list = ['View_from_Second_Link_at_Tuas_to_sg',
+                        'View_from_Second_Link_at_Tuas_to_jh',
+                        'View_from_Tuas_Checkpoint_to_sg',
+                        'View_from_Tuas_Checkpoint_to_jh',
+                        'View_from_Woodlands_Causeway_Towards_Johor_to_sg',
+                        'View_from_Woodlands_Causeway_Towards_Johor_to_jh',
+                        'View_from_Woodlands_Checkpoint_Towards_BKE_to_sg',
+                        'View_from_Woodlands_Checkpoint_Towards_BKE_to_jh']
+
+    counts_df_filter_views = counts_df[counts_df['view'].isin(filtered_views_list)]
+    counts_df_filter_views['date'] = pd.to_datetime(counts_df_filter_views['date'])
+    counts_df_filter_views['day_of_week'] = counts_df_filter_views['date'].dt.day_of_week
+    date_view_group = counts_df_filter_views.groupby(by=['view', 'day_of_week']).mean()
+    date_view_group = date_view_group.reset_index()