Replies to This Discussion

Permalink Reply by Amy on January 26, 2010 at 11:39pm

You might be interested in this:

The third Grenander's problem consists of the estimation of a convex domain when n inside and m outside observations are available. This problem was proposed by Grenander in 1973. There are many applications : estimation of cancerous zones by scanner, etc.

Application to oil industry: when drilling oil wells to determine the location of an oil field, you want to drill as few wells as possible, yet be able to estimate the shape (assumed to be convex) of the oil field as accurately as possible. That's where this data mining technology becomes useful.

• ▶ Reply

Permalink Reply by Vladimir Rastunkov on March 12, 2010 at 12:33am

I can provide you with the following example:

In the oil/gas extraction industry an optimization of the reliability of the equipment plays the crucial role. Let us assume that the system “Well - Electrical Submersible Centrifugal Pump (ESP)” consists of several parts: Pump-compressor pipe, Pump, Gas separator, Seal section, Motor, Power cable. If the operating time of the motor is very short compared to the pump then your system is far from its optimum: nobody will use the pump for the second time and all the equipment will be changed simultaneously. So you should model the reliability of the whole system using the historical data and optimize it. The use of the DM solution is essential since the historical data has a great volume (at list in a large company). Of course the problem requires some specialized automated solution and is not solved with the standard DM instruments. Within the custom programming, system integration and consulting project we have developed such solution using StatSoft STATISTICA system.

• ▶ Reply