feat: small refactor

data_platform/assets/ml/discord_alerts.py

@@ -2,6 +2,7 @@
 
 from pathlib import Path
 
+import pandas as pd
 import requests
 from dagster import (
     AssetExecutionContext,
@@ -51,6 +52,7 @@ def _build_embed(row) -> dict:
     deps=["elo_inference"],
     group_name="ml",
     kinds={"python", "discord"},
+    tags={"manual": "true"},
     description=(
         "Send a Discord notification for newly scored listings whose "
         "predicted ELO exceeds a configurable threshold."
@@ -69,7 +71,7 @@ def listing_alert(
         conn.execute(text(render_sql(_SQL_DIR, "ensure_notified_table.sql")))
 
     query = render_sql(_SQL_DIR, "select_top_predictions.sql")
-    df = __import__("pandas").read_sql(
+    df = pd.read_sql(
         text(query),
         engine,
         params={"min_elo": config.min_elo},

data_platform/assets/ml/elo_inference.py

@@ -16,7 +16,7 @@ from sqlalchemy import text
 
 from data_platform.assets.ml.elo_model import (
     ALL_FEATURES,
-    _preprocess,
+    preprocess,
 )
 from data_platform.helpers import render_sql
 from data_platform.resources import MLflowResource, PostgresResource
@@ -104,7 +104,7 @@ def elo_inference(
     model = mlflow.lightgbm.load_model(model_uri)
 
     # Preprocess features identically to training
-    df = _preprocess(df)
+    df = preprocess(df)
     X = df[ALL_FEATURES].copy()
 
     # Predict normalised ELO and convert back to original scale

data_platform/assets/ml/elo_model.py

@@ -84,7 +84,7 @@ class EloModelConfig(Config):
     mlflow_experiment: str = "elo-rating-prediction"
 
 
-def _preprocess(df: pd.DataFrame) -> pd.DataFrame:
+def preprocess(df: pd.DataFrame) -> pd.DataFrame:
     """Convert raw columns to model-ready numeric features."""
     df["energy_label_num"] = (
         df["energy_label"]
@@ -139,7 +139,7 @@ def elo_prediction_model(
         )
 
     # Preprocess and normalise ELO target
-    df = _preprocess(df)
+    df = preprocess(df)
     df["elo_norm"] = (df["elo_rating"] - 1500) / 100
 
     X = df[ALL_FEATURES].copy()