Adam Gartenberg's Blog

Business Analytics and Optimization, IBM and Social Marketing

Modeling Trains with SPSS

An article in this month's IBM Data Management Magazine (Modeling Trains with SPSS) has a very apt subtitle:  "How sequential analysis can predict the future, and why DB2 might be smarter than you think."

In plain English, the authors looked at how IBM SPSS Modeler and IBM DB2 can analyze data generated by the trains - in this case, over 1000 different types of mechanical and electrical events recorded by the train - to determine which might be early indicators that the train may break down unexpectedly.  To do this task they used sequential analysis to detect patterns in the data.  (This same type of analysis might be used by a retailer to determine that a customer that buys a certain combination of products might be likely to buy a specific third product, as well.)

From the article:

By their nature, trains invite imagination. They conjure visions of exotic travel and long-distance commerce. But trains are also large, complicated, and expensive to repair when they break unexpectedly. For a railway operator, the ability to predict a failure before it happens would be tremendously valuable.
On a modern train, more than 1,000 different types of mechanical and electrical events are recorded, including operational events such as “opening door” or “train is braking,” warning events such as “line voltage frequency out of range” or “compression is low in compressor X,” and failure events such as “pantograph is out of order” or “inverter lockout.” More than 2,000 events are collected on each train every day.

IBM proposed to use sequential analysis to determine whether any of these events or event sequences indicate that a more severe problem is about to occur.

> More