Upload 26 files

app.py

@@ -79,7 +79,7 @@ def main():
             input_save = st.checkbox(config['SAVE_CHECKBOX_TEXT'])
             confirm_params = {
                     'input_save':input_save,
-                    'send_mail':st.session_state["send_mail"],
+                    'send_mail':do_mail_sending,
                     'email_adress': st.session_state["text_input"],
                     'forecast_start_date':st.session_state["input_date"][0].strftime("%Y/%m/%d"),
                     'forecast_end_date':st.session_state["input_date"][1].strftime("%Y/%m/%d")

config.py

@@ -1,3 +1,5 @@
+# tahmin çıktılarında date kolonu sadece yyyy/mm/dd şeklinde yap
+
 from dotenv import load_dotenv
 import os
 

maps/date_models_test/fixed_columns.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:651ae15f3372167ecb195e2419b20b74d08179ea0376a2b5fcc75c31b84ad788
-size 15387
+oid sha256:b157854e71f088b326d9b7a8cca4c23c2d2be8d29c6a34b84247036cbc35b915
+size 14494

maps/date_models_test/shift_features.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4cce186c59cdf90f2a5cb6195c39f98e5924eb55fd88c8f0361b462d3bd7002
+size 6894627

maps/date_models_test/unique_products.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:db7abab3fa1b11bb0121bdc8bbd68d8cbe1a72f4fa95fde94f5bcae0e6296580
-size 9900
+oid sha256:f6566704cf5202f8b6cce4d6aa1d9ab8241056dab9b4089312ccf1c4404bf4d1
+size 9184

models/date_models_test/CATBOOST_FOLD_1.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6a16f83afbb2ccc5acd334d9a5309c20893c670140a053509f3bc9eb6670e32e
-size 3182396
+oid sha256:359f951663f6be9fd14557cae7aa8b33e63dbd0069528322473b82d167ddbe07
+size 5168328

models/date_models_test/CATBOOST_FOLD_2.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ad6e6b32d58e93218fcda2549e1d0a0c43126c17507ab04bf53d24fcf40a2a47
-size 3244499
+oid sha256:646a2e501cdf4a9338587e1046240b9c059f7d5c0e7349aac1358460de74e05f
+size 4778571

models/date_models_test/CATBOOST_FOLD_3.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e19431438d21f520e98d49896e0eaed65f95bde35fabc64b119d139fe4364c6f
-size 3149171
+oid sha256:5addc4fd7536e8a2cdc9d126ea9852ac12dad193aca2ab4bc6bf2e01d7c9ed95
+size 5385151

models/date_models_test/CATBOOST_FOLD_4.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:655eef1028200c63ff8cfcdc0e2ebd752488b20646479f6ca7969eb445a29b3a
-size 2611532
+oid sha256:8d75b9a3e480d90945cfe0d6310759c4bb36612b337db94a69802998d40794d5
+size 241036

models/date_models_test/CATBOOST_FOLD_5.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f9b3e6d1095af0877fcb0096b46eb383597bc25f2116be1401068f3efa11ccdb
-size 227821
+oid sha256:afd6bc4b286114d33acdbecbb2ba04cc846c0fb2f4c7154e4fd1e08dc4ce3f55
+size 77265

pipeline.py

@@ -27,7 +27,10 @@ def run(forecast_start_date, forecast_end_date):
     
     dataframe = prepare_data(
         dataframe=pd.concat([generated_test], axis=0),
-        add_datetime_features=True
+        data=None,
+        split_local_test=config['split_local_test'],
+        add_datetime_features=True,
+        add_lag_features=True
     )
 
     dataframe[config['target']] = np.nan

src/data/load_data.py

@@ -69,7 +69,7 @@ def get_data(
     y_test = generated_test[target]
     generated_test.drop(target, axis=1, inplace=True)
 
-    return generated_train, generated_test, y_test
+    return dataframe, generated_train, generated_test, y_test
 
 def split_train_test(dataframe):
 

src/features/build_features.py

@@ -1,6 +1,15 @@
+import pandas as pd
+from src.utils.helper_functions import save_parquet, load_parquet
+from config import Config
+
+config = vars(Config)
+
 def prepare_data(
         dataframe,
-        add_datetime_features=True
+        data,
+        split_local_test,
+        add_datetime_features=True,
+        add_lag_features=True
     ):
 
     print('Building features...')
@@ -8,6 +17,38 @@ def prepare_data(
     if add_datetime_features:
         dataframe = datetime_features(dataframe)
 
+    if add_lag_features:
+        dataframe = lag_features(dataframe, data, split_local_test)
+
+
+    return dataframe
+
+def lag_features(dataframe, data, split_local_test):
+
+    if split_local_test:
+
+        backlog_cols = [col for col in data.columns if col.endswith('_backlog')]
+
+        lag_backlog_cols = []
+        for col in backlog_cols:
+            for shift in range(9,13,1):
+                
+                shift_col_name = f'{col}_shift_{shift}'
+                data.loc[:, shift_col_name] = data.groupby('product_id')[col].shift(shift)
+                
+                lag_backlog_cols.append(shift_col_name)
+
+        save_parquet(
+            dataframe=data[lag_backlog_cols + ['product_id','date']],
+            path=f'{config["fold_input_directory"]}/shift_features.parquet'
+        )
+
+        map_data = data[lag_backlog_cols + ['product_id','date']]
+    else:
+        map_data = load_parquet(f'{config["fold_input_directory"]}/shift_features.parquet')
+
+    dataframe = pd.merge(dataframe, map_data, how='left', on=['product_id','date'])
+
     return dataframe
 
 def datetime_features(dataframe, date='date', suffix=''):