My experience doing Business Intelligence and Decision Making software.
Optimal decision making requires knowledge and models extracted from huge amount of up-to-date information in real time, and fast algorithms to find optimal solutions using performance functions.
I have strong experience developing algorithms to support Business intelligence�s implementation in complex systems. My algorithms are in successful business intelligence core projects extracting and storing knowledge from large data warehouses, and supporting decision making process in real time.
I have worked for water industry developing risk based capital maintenance planning (Tynemarch Systems Engineering Limited), job submission system for UK e-Science project (Imperial College and University of Manchester), pattern recognition from medical data mining (Brunel University), and several industrial optimisation projects in transport, steel sector, and sugar and alcohol bio industry (South Bank, Gentech Inf, Zillo-Lorenzetti, Procontrol, and Sao Paulo Underground). Some of them required parallel distributed architecture using grid computing and clusters.
I have developed a RAD framework that allows me work with a diversity of adaptive approaches such as genetic algorithms (GA), genetic programming (GP), and statistical adaptation, with fast results and minor disruption in production. I manage my projects (average time 1 year each) with the following milestones:
|
Task |
Goals |
Deliverables |
|
Problem definition |
Understand
user�s problems, company�s system interface, and goals |
Functional specification |
|
Problem�s data acquisition |
Data
description, sources, interdependencies and relationship |
System Meta data |
|
Data Modelling |
Optimal
data storage that cope with optimization model access |
Database
definition, population procedures,�
access interface and data entry system |
|
Data acquisition |
Database
population with operational scenarios for optimization software test |
Test environment and analysis infrastructure |
|
Develop optimisation model |
Optimisation options and prototype development (RAD) |
Optimisation RAD prototype |
|
Model test |
Run prototype with testbed |
Prototype validation |
|
Alfa test |
Run prototype with scenarios |
Comparative analysis optimisation prototype �
prototype improvements and updates |
|
Develop product |
Implement
prototype with improvements and updates to obtain production system |
Optimisation system |
|
Beta Test |
Run scenarios tests with production system |
Optimisation system acceptance |
|
Documentation |
Write
user manual, system manual and training material |
User
manual System
manual Training
set |
|
Training |
Apply training for users and system managers |
Users
able to use the system under production condition |
|
Parallel Production (new system and old one) |
Apply
optimization system in parallel with conventional system |
Optimisation system delivered |
|
Production |
Replace conventional system |
The framework is part of my PhD thesis in processes optimisation using the available information to obtain a decision model by genetic programming, and adapt it in real time with genetic algorithm. The experimental setup to test the framework was a signal processing application with 3 TMS320C44 parallel DSP running GP and one TMS320C32 DSP executing the adaptation under real time for acoustic noise cancellation - a decisive test for any control system due its instability and acoustic feedback.