Mining Data from Automobile ECUs

reilly

i've read(see link below) that some automaker (BMW, Ford, Mercdes) mining data from Automobile ECUs (electronic control unit)

but i can't find concrete examples what they mining
does someone know or have a guess what they mining?


link: http://www.automotiveit.eu/bwm-besserer-service-durch-data-mining/produkte-losungen/id-0024949

reilly

no idea?

wessel

The link you provide doesn't say much.

You can use data mining to make a car automagically:
- break to avoid a collision
- park in some tight spot
- blow your face with cold air just before you fall asleep
- anything

Best regards,

Wessel

harris

reilly wrote:
i've read(see link below) that some automaker (BMW, Ford, Mercdes) mining data from Automobile ECUs (electronic control unit) but i can't find concrete examples what they mining does someone know or have a guess what they mining? link: http://www.automotiveit.eu/bwm-besserer-service-durch-data-mining/produkte-losungen/id-0024949

one of generic applications of DM in autoindustry is whether 'driver is alert while driving' which was the topic of a Kaggle prediction competition (sponsored by Ford) this spring where drivers and the vehicle surroundings were constantly surveilled using mounted sensors at 0.2sec intervals to determine alertness, and thereby reduce highway casualties by tired or drunken drivers. the main need of automakers is however to reduce the number of vehicle recalls which take place at their cost by pre-identifying breakdowns either at development stage.

quick translation from "Eingesetzt wird sie auch von der BMW Group. Dort führen die Mitarbeiter aus dem Qualitätsbereich beispielsweise Häufigkeitsauszählungen von Fehlern eines bestimmten Fahrzeugs durch. Eine wichtige Kennzahl zur Messung der Produktqualität ist die „Anzahl Fehler im Gewährleistungszeitraum“. Eine Reduktion dieser Zahl senkt die Kosten zur Behebung von Fehlern und verbessert die Produkte." says they use data mining for this purpose: identifying specific vehicle faults by counting fault frequency at given time intervals, and they report "Dank der Analyse der Kundenwiederholdaten – Informationen, warum Kunden vermehrt Werkstätten aufsuchen -  verbessert  BMW  die RepeatRepairs um 25 Prozent" to have reduced 'repeatrepairs' cost by 25% by implementing DM fault identification tools (from SPSS, now part of IBM portfolio I think ?).

best, Harri