From a49dedba5250d852c4cfa027ccee976c089e85c8 Mon Sep 17 00:00:00 2001
From: agshaathalla <13521027@mahasiswa.itb.ac.id>
Date: Fri, 10 Jan 2025 16:01:52 +0700
Subject: [PATCH] data drift
---
dags/data_drift.py | 25 +++++++++++++++++++++++++
1 file changed, 25 insertions(+)
create mode 100644 dags/data_drift.py
diff --git a/dags/data_drift.py b/dags/data_drift.py
new file mode 100644
index 0000000..ca149d0
--- /dev/null
+++ b/dags/data_drift.py
@@ -0,0 +1,25 @@
+import pandas as pd
+
+def sample_data(input_csv, output_csv, sample_fraction, random_state=None):
+ data = pd.read_csv(input_csv)
+
+ print("Data sebelum sampling:")
+ print(data.info())
+ print(data.head())
+
+ sampled_data = data.sample(frac=sample_fraction, random_state=random_state)
+
+ print("\nData setelah sampling:")
+ print(sampled_data.info())
+ print(sampled_data.head())
+
+ sampled_data.to_csv(output_csv, index=False)
+ print(f"\nHasil sampling disimpan ke {output_csv}")
+
+if __name__ == "__main__":
+ input_csv = "telco_customer_churn.csv"
+ output_csv = "telco_customer_churn_drift.csv"
+ sample_fraction = 0.2
+ random_state = 42
+
+ sample_data(input_csv, output_csv, sample_fraction, random_state)
--
GitLab