Add default scheduler params and auto-generate fallback

.gitignore

@@ -30,3 +30,7 @@ Data/test_verification/
 !Data/README.md
 !pyproject.toml
 !Data/court_data.duckdb
+
+# Bundled baseline parameters for scheduler
+!scheduler/data/defaults/*.csv
+!scheduler/data/defaults/*.json

scheduler/data/config.py

@@ -4,30 +4,107 @@ This module contains all configuration parameters and constants used throughout
 the scheduler implementation.
 """
 
+import argparse
+import subprocess
+import sys
 from pathlib import Path
-from typing import Dict, List
+from typing import Dict, List, Optional
 
 # Project paths
 PROJECT_ROOT = Path(__file__).parent.parent.parent
 REPORTS_DIR = PROJECT_ROOT / "reports" / "figures"
+DEFAULT_PARAMS_DIR = Path(__file__).parent / "defaults"
+RUN_EDA_SCRIPT = PROJECT_ROOT / "src" / "run_eda.py"
 
-# Find the latest versioned output directory
-def get_latest_params_dir() -> Path:
-    """Get the latest versioned parameters directory from EDA outputs."""
+def _discover_latest_report_dir() -> Optional[Path]:
+    """Return the latest versioned report directory if it exists."""
     if not REPORTS_DIR.exists():
-        raise FileNotFoundError(f"Reports directory not found: {REPORTS_DIR}")
-    
+        return None
+
     version_dirs = [d for d in REPORTS_DIR.iterdir() if d.is_dir() and d.name.startswith("v")]
     if not version_dirs:
-        raise FileNotFoundError(f"No versioned directories found in {REPORTS_DIR}")
-    
-    latest_dir = max(version_dirs, key=lambda d: d.stat().st_mtime)
-    params_dir = latest_dir / "params"
-    
-    if not params_dir.exists():
-        params_dir = latest_dir  # Fallback if params/ subdirectory doesn't exist
-    
-    return params_dir
+        return None
+
+    return max(version_dirs, key=lambda d: d.stat().st_mtime)
+
+
+def _try_run_eda() -> None:
+    """Run the EDA pipeline to regenerate parameters."""
+    if not RUN_EDA_SCRIPT.exists():
+        raise FileNotFoundError(
+            f"Unable to regenerate parameters because {RUN_EDA_SCRIPT} is missing. "
+            "Please ensure the EDA pipeline is available."
+        )
+
+    print("No EDA outputs found. Running src/run_eda.py to generate parameters...", file=sys.stderr)
+    result = subprocess.run([sys.executable, str(RUN_EDA_SCRIPT)], check=False)
+    if result.returncode != 0:
+        raise RuntimeError(
+            "Failed to regenerate parameters via src/run_eda.py. "
+            "Check the data dependencies and try again."
+        )
+
+
+# Find the latest versioned output directory
+def get_latest_params_dir(
+    regenerate: bool = False,
+    allow_generate: bool = True,
+    allow_defaults: bool = True,
+    prefer_defaults: bool = False,
+) -> Path:
+    """Get the latest parameters directory from EDA outputs or bundled defaults.
+
+    The lookup strategy is:
+    1) Use the latest versioned directory in reports/figures (unless regenerating).
+    2) Optionally run the EDA pipeline to create parameters when none exist.
+    3) Fallback to bundled defaults when available.
+
+    Args:
+        regenerate: When True, always run the EDA pipeline before resolving params.
+        allow_generate: If True, run EDA automatically when no outputs exist.
+        allow_defaults: If True, fallback to bundled defaults if EDA outputs are missing.
+        prefer_defaults: If True, return bundled defaults immediately when available.
+
+    Returns:
+        Path to a directory containing parameter files.
+
+    Raises:
+        FileNotFoundError: When parameters cannot be located or generated.
+        RuntimeError: When regeneration is attempted but fails.
+    """
+
+    if prefer_defaults and allow_defaults and DEFAULT_PARAMS_DIR.exists():
+        print(
+            "Using bundled baseline parameters from scheduler/data/defaults (preferred).",
+            file=sys.stderr,
+        )
+        return DEFAULT_PARAMS_DIR
+
+    if not regenerate:
+        latest_dir = _discover_latest_report_dir()
+        if latest_dir:
+            params_dir = latest_dir / "params"
+            return params_dir if params_dir.exists() else latest_dir
+
+    if regenerate or (allow_generate and not _discover_latest_report_dir()):
+        _try_run_eda()
+        latest_dir = _discover_latest_report_dir()
+        if latest_dir:
+            params_dir = latest_dir / "params"
+            return params_dir if params_dir.exists() else latest_dir
+
+    if allow_defaults and DEFAULT_PARAMS_DIR.exists():
+        print(
+            "Using bundled baseline parameters from scheduler/data/defaults (EDA outputs not found).",
+            file=sys.stderr,
+        )
+        return DEFAULT_PARAMS_DIR
+
+    missing_reports_msg = (
+        "No parameter directory found. Ensure EDA has been run (python src/run_eda.py) "
+        "or use bundled defaults via get_latest_params_dir(allow_defaults=True)."
+    )
+    raise FileNotFoundError(missing_reports_msg)
 
 # Court operational constants
 WORKING_DAYS_PER_YEAR = 192  # From Karnataka High Court calendar
@@ -120,3 +197,37 @@ RANDOM_SEED = 42
 # Logging configuration
 LOG_LEVEL = "INFO"
 LOG_FORMAT = "%(asctime)s - %(name)s - %(levelname)s - %(message)s"
+
+
+def _parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(
+        description=(
+            "Resolve the scheduler parameter directory, optionally regenerating via the EDA pipeline."
+        )
+    )
+    parser.add_argument(
+        "--regenerate",
+        action="store_true",
+        help="Run src/run_eda.py before resolving parameters.",
+    )
+    parser.add_argument(
+        "--use-defaults",
+        action="store_true",
+        help="Force use of bundled defaults instead of scanning reports/figures.",
+    )
+    return parser.parse_args()
+
+
+def _main() -> None:
+    args = _parse_args()
+    params_dir = get_latest_params_dir(
+        regenerate=args.regenerate,
+        allow_generate=not args.use_defaults,
+        allow_defaults=True,
+        prefer_defaults=args.use_defaults,
+    )
+    print(params_dir)
+
+
+if __name__ == "__main__":
+    _main()

scheduler/data/defaults/adjournment_proxies.csv

@@ -0,0 +1,11 @@
+Remappedstages,casetype,p_adjourn_proxy,p_not_reached_proxy,n
+PRE-ADMISSION,CRP,0.3,0.05,80
+PRE-ADMISSION,CA,0.28,0.05,80
+ADMISSION,CRP,0.32,0.07,90
+ADMISSION,CA,0.35,0.08,90
+FRAMING OF CHARGES,RSA,0.25,0.05,70
+EVIDENCE,RSA,0.4,0.1,100
+ARGUMENTS,RFA,0.22,0.05,85
+INTERLOCUTORY APPLICATION,CCC,0.45,0.1,40
+ORDERS / JUDGMENT,CP,0.18,0.02,60
+SETTLEMENT,CMP,0.15,0.02,30

scheduler/data/defaults/case_type_summary.csv

@@ -0,0 +1,8 @@
+CASE_TYPE,n_cases,disp_median,disp_p90,hear_median,gap_median
+CRP,1200,180,320,6,25
+CA,1100,200,360,7,28
+RSA,900,240,420,8,32
+RFA,850,210,400,7,30
+CCC,500,160,280,5,20
+CP,450,150,260,5,18
+CMP,300,140,240,4,16

scheduler/data/defaults/court_capacity_global.json

@@ -0,0 +1,4 @@
+{
+  "slots_median_global": 151,
+  "slots_p90_global": 210
+}

scheduler/data/defaults/stage_duration.csv

@@ -0,0 +1,12 @@
+STAGE,RUN_MEDIAN_DAYS,RUN_P90_DAYS,HEARINGS_PER_RUN_MED,N_RUNS
+PRE-ADMISSION,7,14,1,150
+ADMISSION,14,30,1,150
+FRAMING OF CHARGES,21,45,2,120
+EVIDENCE,60,120,3,110
+ARGUMENTS,30,60,2,130
+INTERLOCUTORY APPLICATION,14,28,1,60
+SETTLEMENT,10,20,1,50
+ORDERS / JUDGMENT,14,28,1,140
+FINAL DISPOSAL,1,2,1,200
+OTHER,7,14,1,40
+NA,1,1,1,10

scheduler/data/defaults/stage_transition_entropy.csv

@@ -0,0 +1,11 @@
+STAGE_FROM,entropy
+PRE-ADMISSION,0.6
+ADMISSION,0.8
+FRAMING OF CHARGES,0.7
+EVIDENCE,0.5
+ARGUMENTS,0.4
+INTERLOCUTORY APPLICATION,0.65
+ORDERS / JUDGMENT,0.3
+SETTLEMENT,0.1
+OTHER,0.2
+NA,0.0

scheduler/data/defaults/stage_transition_probs.csv

@@ -0,0 +1,20 @@
+STAGE_FROM,STAGE_TO,N,row_n,p
+PRE-ADMISSION,ADMISSION,100,1,0.7
+PRE-ADMISSION,FINAL DISPOSAL,100,2,0.3
+ADMISSION,FRAMING OF CHARGES,120,1,0.6
+ADMISSION,ARGUMENTS,120,2,0.25
+ADMISSION,FINAL DISPOSAL,120,3,0.15
+FRAMING OF CHARGES,EVIDENCE,90,1,0.65
+FRAMING OF CHARGES,ARGUMENTS,90,2,0.2
+FRAMING OF CHARGES,FINAL DISPOSAL,90,3,0.15
+EVIDENCE,ARGUMENTS,110,1,0.7
+EVIDENCE,FINAL DISPOSAL,110,2,0.3
+ARGUMENTS,ORDERS / JUDGMENT,130,1,0.75
+ARGUMENTS,FINAL DISPOSAL,130,2,0.25
+INTERLOCUTORY APPLICATION,ARGUMENTS,30,1,0.6
+INTERLOCUTORY APPLICATION,FINAL DISPOSAL,30,2,0.4
+ORDERS / JUDGMENT,FINAL DISPOSAL,140,1,0.9
+ORDERS / JUDGMENT,SETTLEMENT,140,2,0.1
+SETTLEMENT,FINAL DISPOSAL,50,1,1.0
+OTHER,FINAL DISPOSAL,40,1,1.0
+NA,NA,10,1,1.0