v2: update data_loader.py

data_loader.py

@@ -8,6 +8,7 @@ import pandas as pd
 import streamlit as st
 from huggingface_hub import hf_hub_download
 
+
 DATASET_REPO = "buckeyeguy/osc-usage-data"
 
 
@@ -42,6 +43,24 @@ def load_data() -> tuple[pd.DataFrame, pd.DataFrame, dict]:
     if "walltime_used" in jobs.columns:
         jobs["walltime_hours"] = jobs["walltime_used"] / 3600.0
 
+    # Timeout classification — interactive vs batch
+    if "launch_method" in jobs.columns and "last_state" in jobs.columns:
+        import numpy as np
+
+        from config import INTERACTIVE_METHODS
+
+        is_timeout = jobs["last_state"] == "TIMEOUT"
+        is_interactive = jobs["launch_method"].isin(INTERACTIVE_METHODS)
+        jobs["timeout_category"] = np.where(
+            ~is_timeout,
+            jobs["last_state"],
+            np.where(is_interactive, "Interactive Timeout", "Batch Timeout"),
+        )
+
+    # Queue wait time
+    if "submit_time" in jobs.columns and "start_time" in jobs.columns:
+        jobs["wait_hours"] = (jobs["start_time"] - jobs["submit_time"]).dt.total_seconds() / 3600.0
+
     return jobs, snapshots, metadata