MultiProcessTraceabilityEvents

Item timelines, lead times, parallel processing and routing paths across a station topology with parallel cells, merges and splits. Each station publishes the ID of the item it is working on.

ts_shape.events.production.multi_process_traceability · Production guide · API reference

Example

import pandas as pd
from ts_shape.events.production.multi_process_traceability import MultiProcessTraceabilityEvents

t0 = pd.Timestamp("2024-01-01 06:00")
rows = [(0, "cell_a1:order_id", "ORD-1"), (4, "cell_a1:order_id", "ORD-1"),
        (2, "cell_b1:order_id", "ORD-1"), (6, "cell_b1:order_id", "ORD-1"),  # parallel to A1
        (8, "merge:order_id", "ORD-1"), (12, "merge:order_id", "ORD-1"),
        (5, "cell_a1:order_id", "ORD-2"), (9, "cell_a1:order_id", "ORD-2"),
        (13, "merge:order_id", "ORD-2"), (16, "merge:order_id", "ORD-2")]
df = pd.DataFrame([{"systime": t0 + pd.Timedelta(minutes=m), "uuid": u, "value_string": v}
                   for m, u, v in rows])
tracer = MultiProcessTraceabilityEvents(df, processes=[
    {"id_uuid": "cell_a1:order_id", "station": "Cell A1"},
    {"id_uuid": "cell_b1:order_id", "station": "Cell B1"},
    {"id_uuid": "merge:order_id", "station": "Merge"}])
print(tracer.lead_time()[["item_id", "station_path", "lead_time_seconds", "has_parallel"]])

Methods

Method	Returns
build_timeline(infer_dwell_from_next=False)	one row per item and station visit: item_id, station, id_uuid, start, end, duration_seconds, step_sequence, is_parallel
lead_time()	per item: first_seen, last_seen, lead_time_seconds, stations_visited, station_path, has_parallel
parallel_activity()	overlapping visits: item_id, station_a, station_b, overlap_start, overlap_end, overlap_seconds
handover_log()	one row per handover trigger: timestamp, item_id, from_station, to_station, handover_value
station_statistics(infer_dwell_from_next=False)	per station and id_uuid: item_count, min_/avg_/max_/total_dwell_seconds
routing_paths()	per path: item_count, avg_/min_/max_lead_time_seconds, has_parallel_steps

Notes

• processes maps each station's item-ID signal (id_uuid) to a station name. Parallel cells of one station type can share a name; station_statistics() still separates them by id_uuid. Visit order comes from timestamps; overlapping visits are flagged as parallel.
• Optional handovers=[{"uuid", "from_station", "to_station"}]: a value > 0 (or True) is a transfer, and the item is taken from the from-station's ID signal at that time.
• By default a visit ends at the item's last sample at the station, so a station that logs the ID only on arrival gives a dwell of 0. infer_dwell_from_next=True ends each sequential visit at the item's next station arrival; parallel visits keep their own span.