PM: Transition System Miner

Siap! Berikut script **Transition System Miner** dengan **PM4Py** yang membaca **file XES terlampir** (`/mnt/data/BPIC_2012_A.xes`), menambang **Transition System (TS)**, lalu:

• Cetak ringkasan jumlah **state** dan **transition**
• Simpan **gambar TS** (`transition_system.png`)
• Ekspor daftar **state** (`ts_states.csv`) dan **transition** (`ts_transitions.csv`)

> **Install dulu (sekali saja):** > > ```bash > pip install pm4py pandas graphviz > # pastikan sistem punya Graphviz binary (dot). Linux (Debian/Ubuntu): > # sudo apt-get install graphviz > ```

---

1. 1. 1. `transition_system_miner.py`

```python

1. !/usr/bin/env python3
2. -*- coding: utf-8 -*-

import argparse import sys from pathlib import Path import pandas as pd

from pm4py.objects.log.importer.xes import importer as xes_importer from pm4py.algo.discovery.transition_system import algorithm as ts_discovery from pm4py.visualization.transition_system import visualizer as ts_visualizer

def discover_transition_system(xes_path: Path, image_out: Path, states_out: Path, trans_out: Path):

   # 1) Load event log
   log = xes_importer.apply(str(xes_path))

   # 2) Discover Transition System
   # Parameter kunci (ubah jika perlu)
   parameters = {
       "case_glue": "case:concept:name",
       "activity_key": "concept:name",
       "timestamp_key": "time:timestamp",
   }
   ts = ts_discovery.apply(log, parameters=parameters)

   # 3) Visualisasi & simpan ke file
   gviz = ts_visualizer.apply(ts)
   ts_visualizer.save(gviz, str(image_out))

   # 4) Ekspor states & transitions ke CSV
   # Struktur objek TS di PM4Py: ts.states (set of State), ts.transitions (set of Transition)
   # State biasanya memiliki 'name' (id) dan 'label' (representasi state)
   states_rows = []
   for s in ts.states:
       sid = getattr(s, "name", None)
       slabel = getattr(s, "label", None)
       # fallback agar tetap terisi
       if sid is None:
           sid = str(s)
       if slabel is None:
           slabel = str(s)
       states_rows.append({"state_id": sid, "state_label": slabel})

   trans_rows = []
   for t in ts.transitions:
       # transition memiliki source (from), target (to), label (activity/event class)
       src = getattr(t, "from_state", getattr(t, "from", None))
       dst = getattr(t, "to_state", getattr(t, "to", None))
       lab = getattr(t, "label", None)

       # Ambil id/label state sumber & tujuan
       def state_id_label(state_obj):
           if state_obj is None:
               return None, None
           sid = getattr(state_obj, "name", None) or str(state_obj)
           slb = getattr(state_obj, "label", None) or str(state_obj)
           return sid, slb

       src_id, src_label = state_id_label(src)
       dst_id, dst_label = state_id_label(dst)

       trans_rows.append({
           "source_id": src_id,
           "source_label": src_label,
           "target_id": dst_id,
           "target_label": dst_label,
           "transition_label": lab if lab is not None else ""
       })

   pd.DataFrame(states_rows).to_csv(states_out, index=False)
   pd.DataFrame(trans_rows).to_csv(trans_out, index=False)

   # 5) Ringkasan
   print("=== Transition System Summary ===")
   print(f"States     : {len(states_rows)}")
   print(f"Transitions: {len(trans_rows)}")
   print(f"Gambar     : {image_out}")
   print(f"States CSV : {states_out}")
   print(f"Trans CSV  : {trans_out}")

def main():

   ap = argparse.ArgumentParser(description="Transition System Miner using PM4Py (from XES)")
   ap.add_argument("xes_path", type=str, help="Path ke file .xes")
   ap.add_argument("--img", type=str, default="transition_system.png", help="Output image (PNG)")
   ap.add_argument("--states_csv", type=str, default="ts_states.csv", help="Output CSV daftar state")
   ap.add_argument("--trans_csv", type=str, default="ts_transitions.csv", help="Output CSV daftar transition")
   args = ap.parse_args()

   xes_path = Path(args.xes_path)
   if not xes_path.exists():
       print(f"[ERROR] File tidak ditemukan: {xes_path}", file=sys.stderr)
       sys.exit(1)

   discover_transition_system(
       xes_path=xes_path,
       image_out=Path(args.img),
       states_out=Path(args.states_csv),
       trans_out=Path(args.trans_csv),
   )

if __name__ == "__main__":

   main()

```

---

1. 1. 1. Cara menjalankan (pakai file terlampir)

```bash python transition_system_miner.py /mnt/data/BPIC_2012_A.xes \

 --img ts_BPICA.png \
 --states_csv ts_BPICA_states.csv \
 --trans_csv ts_BPICA_transitions.csv

```

---

1. 1. 1. Opsi & catatan

• **Kolom kunci** diset ke standar PM4Py:

 `case:concept:name`, `concept:name`, `time:timestamp`.
 Jika log Anda memakai nama kolom berbeda, ubah di `parameters`.

• **Graphviz** diperlukan agar file PNG bisa disimpan. Jika belum ada, install `graphviz` (OS) selain paket Python-nya.
• Untuk log besar, TS bisa sangat besar. Anda bisa mulai dari subset (filtering case/variant) sebelum menambang TS:

 * Filter variant Top-K, atau
 * Filter rentang tanggal tertentu.

• Jika ingin **lihat** langsung (open viewer), ganti `ts_visualizer.save(...)` menjadi `ts_visualizer.view(gviz)` (akan membuka jendela viewer apabila environment mendukung).

Mau saya tambahkan versi dengan **filter Top-K variants** atau **prefix-length limit** biar graf tidak terlalu padat?