Case Studies

Case 1
Powerful Predictive Data Analytics Strategies
to Lower the Risk of Underwriting Loans

The Client's Challenge

ZC, a financial service company, used FICO and other industry standard risk scores to underwrite its loans, but the default rate was only a little better than the industry average.

Our Solution

We combined thousands of variables into many meta variables with latent variable modeling, then built few predictive models with meta variables and ensemble all of the models for final scoring.

The Impact

ZC is able to cut default rate by half while maintaining the same approval rate.

Case 1
Powerful Predictive Data Analytics Strategies to Lower the Risk of Underwriting Loans

The Client's Challenge

ZC, a financial service company, used FICO and other industry standard risk scores to underwrite its loans, but the default rate was only a little better than the industry average.

Our Solution

We combined thousands of variables into many meta variables with latent variable modeling, then built few predictive models with meta variables and ensemble all of the models for final scoring.

The Impact

ZC is able to cut default rate by half while maintaining the same approval rate.

Case 2
Better Click-Through Rates Prediction Model to Increase Advertising Revenue

The Client's Challenge

SZ, an internet search company, did not have a good predictive model of its click-through rate on products offered. This produced non-ideal web rankings and low advertising revenue.

Our Solution

We used time series modeling with Bayesian smoothing and built models in parallel for more than two thousand categories.

The Impact

SZ has more accurate prediction of its click-through rates, which leads to better web page rankings and a 30% increase in advertising revenue.

Case 3
Advanced Propensity Model to Increase Marketing Effectiveness

The Client's Challenge

IMG, a big distributor of thousands of computer and electronic products, used a propensity model to determine which leads to be contacted. However, this propensity model performed poorly as leads selected by this model did not increase response rate as much.

Our Solution

We merged the old propensity model with more data sources, improved modeling processes and used advanced analytics to deal with missing values.

The Impact

IMG’s Marketing Department has better leads, which increased marketing effectiveness and ROI by 50%.

Case 4
Advanced Causal Modeling to Optimize Marketing Spending

The Client's Challenge

Sears, a major retailer in the US, received conflicting proposals on marketing spending to increase sales.

Our Solution

Using survey data, we built advanced causal model of consumer purchases with the total effects of various factors on final sales calculated for optimization.

The Impact

Sears is able to obtain an optimal combination of marketing programs with a sales increase of 80% over a year period for more than 10 testing stores.

Interested with our works ?

Contact Us

Get In Touch

The Royal 55 Office Tower #201, Surabaya 60215
hello@narasiodata.com
(62) 819696699

Copyright © 2020 RMDS Narasio Data | Powered by RMDS Narasio Data