Breakthrough Argyle Data™ Application Successfully Predicts Mobile Subscriber Creditworthiness in Multiple Operator Trials

Print Friendly, PDF & Email

AI/machine learning company Argyle Data™ has successfully concluded a series of trials with European and Latin American operators, using new algorithms and neural network architectures analyzing real carrier data to accurately predict subscribers’ intention and ability to pay monthly service bills.

Our solution is a quantum leap in subscriber validation.  It not only identifies undesirable subscribers at the time of signing on for a mobile service, but also successfully predicts delinquencies in the following 60 – 90 days. This breakthrough is demonstrably more accurate than all other approaches and redefines the cost model for subscriber credit checking,” said Vikash Varma, President and CEO at Argyle Data.

Approximately 40% of all operator bad debts result from subscription fraud and default. Existing credit rating and analytics systems have not been able to identify high risk applicants from genuine subscribers. Even with limited data sets during very short time frames, Argyle Data achieved over 70% accuracy in identifying defaulters – an unprecedented result in reducing subscriber-related losses.

This solution is built upon Argyle Data’s existing machine learning (ML) applications that predict subscription fraud, stolen or false identities, SIM fraud, dealer scams, and handset theft schemes categorized as ‘Never Pay’

In this application, Argyle Data’s models are designed specifically to provide mobile carriers with early insight into future credit and payment issues by predicting the financial ability and future payment profile of subscribers. In the recent trials, the models were focused on predicting payment default sixty and ninety days out.

Creating the features requires a combination of mobile domain expertise and Data Scientist background to extract the relevant information from the data,” said Padraig Stapleton, VP Engineering at Argyle Data.  “This allows our models to predict the way that subscribers’ behavior will change and accurately identify those individuals with a high likelihood of payment issues in the future.”

Building upon this experience, Argyle Data has added a layer of processing using a neural network architecture with more focus on model tuning and less on feature engineering. Leveraging these different but complementary ML approaches, Argyle Data is gaining better results and more transparency on the outcome of the predictive processes.

With historical data provided by the carrier, the Argyle Data solution is able to predict with a high degree of accuracy any subscriber payment issues likely to occur as far ahead as two to three months before default.

Machine learning analytics is increasingly viewed as a key tool to improve the efficiency and profitability of mobile carrier networks [1]. Industry research firm International Data Corporation (IDC) forecasts worldwide spending on cognitive and artificial intelligence (AI) systems [2] will increase by 59.3% to $12.7 billion by the end of 2017 and reach $46 billion in 2020.

Argyle Data is currently collaborating with major credit rating companies and systems integrators on subscriber validation trials with leading edge carriers in Europe and Latin America.




Sign up for the free insideBIGDATA newsletter.

Speak Your Mind