All-Time Parts Prediction (ATP) Demo

All-Time Parts Prediction (ATP) Demo

What is ATP about and which value does it bring?

ATP predicts the demand for service parts (especially in the automotive industry) for so-called long-time-buy or all-time-buy decisions. This future demand may cover the next 10-20 years and is difficult to estimate, which often leads to buying way too much. Consequently, after many years of sitting in the warehouse, at the end, huge amounts need to be scrapped. This causes high inventory and warehousing costs, which can be significantly reduced by more accurate demand predictions. The IBM ATP Solution has been developed to do exactly that: to predict all-time demand with high accuracy.

How does ATP work?

The ATP solution uses a combination of various prediction models to account for the different demand patterns of service parts (fast movers, slow movers, non-movers): Parts with full historic life cycles are used to predict the demand of parts which are in the middle of their life cycle. For slow movers, a neural network approach is used. For new parts, a combination of clustering and decision trees is used to identify the most likely form of their life cycle. The ATP solution was developed using IBM SPSS Modeler, augmented by R, PERL and Jython scripts for data and results preparation purposes.

Ongoing and envisioned ATP research

In cooperation with the Karlsruhe Service Research Institute, the ATP is continuously developed further. Currently, two master theses are being written on the topics of “Planning the life-cycle demand of new parts” and “Predicting the future demand with vehicle production data”. Future research will, on the one hand, concentrate on refining the existing prediction models, and on the other hand around the development of further causal based prediction models – in particular, such models that take known failure rates into account.


There are several reasons why implementing ATP on SDIL is of advantage to all participants.

  1. The ATP solution has been proven successful for several automotive companies and therefore the ATP Demo shows a realistic Big Data use case.
  2. The data used for the ATP Demo are anonymous real data which are normally difficult to get.
  3. The ATP Demo will show that the ATP Solution is cloud ready.
  4. For further research, the ATP Demo can be augmented with new models and used as a test environment for such models.
  5. The ATP Demo will demonstrate the capabilities of building an analytics supply chain using IBM SPSS Modeler in combination with IBM SPSS CADS, Jython and R components.
  6. The ATP Demo comes with its own data and – once implemented and tested on SDIL – will be ready for use.
  7. The ATP Demo can be used to attract automotive clients to participate in SDIL.

Performance Measurement

  • Successful implementation of the ATP Demo as described above
  • Number of times the ATP Demo is shown to a potential customer
  • Number of times that ATP led to the successful acquisition of a new project
  • Winning new SDIL participants due to the ATP Demo

Data Innovation Community

Industry 4.0

Project Partners

IBM Deutschland GmbH, Karlsruhe Service Research Institute (KSRI)

Contact Person

Dr. Peter Korevaar,

Project Duration

Jan. 2017 – Jan. 2019