LatentView Analytics Takes Top Honors At The PAKDD 2010 Data Mining Competition

LatentView teams were recognized as joint Runners-up at the 14th edition of the prestigious competition. The challenge was to re-calibrate a credit risk assessment system based on biased data.

Jun 24, 2010 – New York, NY / Chennai, India

LatentView Analytics (LatentView), a leading Predictive Analytics and Decision Management services firm, was ranked as the top analytical services firm in the recently concluded 14th Pacific-Asia Knowledge Discovery and Data Mining conference (PAKDD 2010) data mining competition. LatentView teams jointly finished at 2nd position to take top honors at the 14th edition of the prestigious event, which saw over 200 registrations from all continents producing over 20,000 page views from over 1,100 people. The LatentView team made a 20 minute presentation on the analytical modeling approach adopted at the PAKDD conference held at Hyderabad, India on June 24, 2010.

“The continued success at the PAKDD data mining competition is testament to the creativity in problem solving, rigor in approach and quest for excellence in everything that we do. Coming on the heels of our recognition by Deloitte as one of Asia’s Fastest Growing Technology Companies, this success validates that our recent achievements are a result of our systematic approach, robust methodology and stringent quality processes”, says VenkatViswanathan, CEO of LatentView Analytics.

The challenge was to develop a predictive risk assessment model to rank order credit card applicants on their probability to default on payments. The teams used 4 classes of techniques to develop the predictive model – Regression, Decision Tree, Gradient Boosting Algorithm and Naïve Bayes. The key differentiator was in the innovative approach which involved developing an Ensemble of Models (ANN or Artificial Neural Networks approach) using the techniques listed on various random samples, refining the scores using the Weight of Evidence approach, and using novel methods to combine the individual predictions to determine the final prediction.

“Participating in PAKDD 2010 has been an exciting experience, and to be ranked higher than teams from industry leaders like Fair Issac is a tremendous achievement. The problem was challenging as the model had to overcome the sample bias in the development sample and still be robust enough to deliver accurate results over time. It was an exciting mix of advanced statistics, machine learning algorithms, data handling and domain knowledge” says PriyaBalakrishnan, a member of the winning team.

About PAKDD 2010

The 14th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD 2010) is a leading international conference in the areas of data mining and knowledge discovery. It provides an international forum for researchers and industry practitioners to share their new ideas, original research results and practical development experiences from all KDD related areas including data mining, data warehousing, machine learning, databases, statistics, knowledge acquisition and automatic scientific discovery, data visualization, causal induction and knowledge-based system.

Leave a Reply

Scroll to Top