Create train_local.py

train_local.py

@@ -0,0 +1,38 @@
+import pandas as pd
+import numpy as np
+import xgboost as xgb
+import joblib
+
+# 1. Download the dataset
+url = "https://raw.githubusercontent.com/datasets-machine-learning/nasa-turbofan-failure-prediction/master/data/train_FD001.txt"
+cols = ['unit', 'cycles', 'os1', 'os2', 'os3'] + [f's{i}' for i in range(1, 22)]
+df = pd.read_csv(url, sep='\s+', header=None, names=cols)
+
+# 2. Calculate Remaining Useful Life (RUL) - This is our 'Target'
+# We find the maximum cycle for each engine and subtract the current cycle
+max_cycles = df.groupby('unit')['cycles'].max().reset_index()
+max_cycles.columns = ['unit', 'max_of_unit']
+df = df.merge(max_cycles, on='unit', how='left')
+df['RUL'] = df['max_of_unit'] - df['cycles']
+
+# 3. Feature Selection
+# We use the most important sensors for a jet engine
+features = ['cycles', 's2', 's3', 's4', 's7', 's8', 's11', 's12', 's13', 's15', 's17', 's20', 's21']
+X = df[features]
+y = df['RUL']
+
+# 4. Train the Model (XGBoost)
+print("Training the model... please wait.")
+model = xgb.XGBRegressor(
+    n_estimators=100, 
+    learning_rate=0.1, 
+    max_depth=5, 
+    objective='reg:squarederror'
+)
+model.fit(X, y)
+
+# 5. Save the 'Brain' of the AI
+joblib.dump(model, 'engine_model.pkl')
+
+print("✅ Success! 'engine_model.pkl' has been created in your folder.")
+print("Now, upload this file to your Hugging Face Space.")