- Products & Services
- Knowledge Base
MINSK, BELARUS, April 21, 2017 /24-7PressRelease/ -- An AI-powered scoring startup GiniMachine announced the launch of an artificially intelligent platform, which provides faster and more accurate credit scoring results.
"The majority of lenders still use the same traditional credit scoring approach. In many cases it is inaccurate, slow, requires tonnes of manual work and access to expensive software," says Dmitry Dolgorukov, GiniMachine's CEO and Co-Founder.
"Getting the hang of lending best practices helped us develop an AI-based credit scoring solution. A platform that mimics the thought process of a top risk manager, but with the accuracy and speed of a machine," continues Dmitry.
"What makes us stand out from the similar industry solutions? It's easy. We don't use any generic models or pre-defined segments that suit everyone and no one. Our robot works solely with the client's data and creates custom scoring models that 100% meet their business needs. This is game changing," sums up Dmitry.
What Makes GiniMachine Unique
With a Gini coefficient of 0.68 and higher, GiniMachine scoring robot shows the results comparable to those achieved by the top risk management professionals.
GiniMachine automatically learns from the lender's customer database, creates a unique credit scoring model and then makes loan predictions.
The system uses a 3-stage pipeline:
1. Data Origination. The robot requires approximately 1.000 records of previously issued loans.
2. Robot Training. GiniMachine automatically analyses all the data and builds a unique scoring model that suits lender's business needs.
3. Full-scale Operation. The system instantly analyses whether credit applicants will pay back their loans or not.
Founded in 2016, GiniMachine is a fast-growing Fintech startup with the focus on credit scoring. The system is based on the best machine learning algorithms and automatically creates advanced scoring models using lenders' data insights. For more information see http://www.ginimachine.com
# # #