dates model

2026-08-19 04:14:00 +00:00 · 2023-08-02 15:32:23 +00:00
parent f8c337e731
commit 9bc6c74ad1
1 changed files with 80 additions and 0 deletions
@@ -0,0 +1,80 @@
+# Python script to compute advance working dates
+
+# Weekend Example: using Saturday or Sunday and add 2 working days:
+# Monday = 1 working day, Tuesday = 1 working day, Wednesday = Delivery Date
+# Assumption for weekends, base date is NOT considered as one working day
+
+# Weekday Example: using Monday and adding 2 working days:
+# Monday = 1 working day, Tuesday = 1 working day, Wednesday = Delivery Date
+# Assumption for weekdays, base date is considered as one working day
+
+# Forecasting on future dates without information on holidays will not be accurate.
+
+
+import pandas as pd
+from datetime import datetime as dt, timedelta
+
+
+# Function to check if a given date is a weekend [Saturday(5) or Sunday(6)]
+def is_weekend(date):
+    return (date.weekday() == 5) | (date.weekday() == 6)
+
+
+# Function to compute the in advance working days
+def compute_working_days(start_date, num_working_days, holiday_list, after_cut_off = 0):
+    # after_cut_off = 0 for orders before 4pm
+    # after_cut_off = 1 for orders after 4pm
+
+    counter = 0
+
+    # For dates on weekend or holiday or after cut off time, additional 1 working day to the loop
+    if (after_cut_off == 1) | is_weekend(start_date) | (start_date in holiday_list):
+        counter -= 1
+
+    # Loop to increase n number of working days, if weekend/holiday, skip counter
+    while counter < num_working_days:
+        start_date = start_date + timedelta(days=1)
+
+        if is_weekend(start_date) | (start_date in holiday_list):
+            continue
+
+        counter += 1
+
+    return start_date
+
+
+# Main function
+def model(dbt, session):
+    
+    # Setting configuration
+    dbt.config(materialized="table",
+                packages = ["pandas"])
+
+    # Import data from upstream dbt model
+    df_date = dbt.ref("int__dates")
+
+    # Extract holiday dates into a list
+    df_work = df_date.to_pandas()
+    df_new = df_work[df_work["IS_COMPANY_HOLIDAY"] == 1]
+    df_holiday_list = df_new["DATE_DAY"].tolist()
+
+    # Apply function to df
+    df_work["ADD_1_WORKING_DAY_INCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 1, df_holiday_list, 0), axis=1)
+    df_work["ADD_2_WORKING_DAY_INCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 2, df_holiday_list, 0), axis=1)
+    df_work["ADD_3_WORKING_DAY_INCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 3, df_holiday_list, 0), axis=1)
+    df_work["ADD_5_WORKING_DAY_INCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 5, df_holiday_list, 0), axis=1)
+    df_work["ADD_30_WORKING_DAY_INCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 30, df_holiday_list, 0), axis=1)
+    df_work["ADD_90_WORKING_DAY_INCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 90, df_holiday_list, 0), axis=1)
+    df_work["ADD_365_WORKING_DAY_INCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 365, df_holiday_list, 0), axis=1)
+
+    # For orders after cut off time 4pm
+    # Affects only the weekdays
+    df_work["ADD_1_WORKING_DAY_EXCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 1, df_holiday_list, 1), axis=1)
+    df_work["ADD_2_WORKING_DAY_EXCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 2, df_holiday_list, 1), axis=1)
+    df_work["ADD_3_WORKING_DAY_EXCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 3, df_holiday_list, 1), axis=1)
+    df_work["ADD_5_WORKING_DAY_EXCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 5, df_holiday_list, 1), axis=1)
+    df_work["ADD_30_WORKING_DAY_EXCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 30, df_holiday_list, 1), axis=1)
+    df_work["ADD_90_WORKING_DAY_EXCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 90, df_holiday_list, 1), axis=1)
+    df_work["ADD_365_WORKING_DAY_EXCLUDED_BASE_DATE"] = df_work.apply(lambda row: compute_working_days(row["DATE_DAY"], 365, df_holiday_list, 1), axis=1)
+
+    return df_work