Headquarters: Nonthaburi, Thailand
- Modern Data Platform: Cloudera Enterprise
- Workloads: Data Warehouse, Data Science & Engineering, Operational Database
- Key Components: Apache HBase, Apache Hive, Apache Impala, Apache Spark
- BI & Analytics Tool: QlikView, Qlik Sense, Microstrategy, SAS, Jasper, RStudio, Python, Cloudera Data Science Workbench
- Predictive analytics
- Machine learning
- Credit approvals
- Regulatory compliance
- Transactional data
- Social media data
- Third party data
- Improves customer experience with reduction in time to access statements from up to a week to seconds
- Estimated to save the bank US$5 million over five years
- More accurately assesses credit risk
- Decreases loan approval process time
- Reduces the number of false positives and better pinpoints actual fraud cases for improved fraud detection
Big data scale
- 35 years of data with approximately two billion transactions added annually
Kasikorn Bank is creating the digital bank of the future with Cloudera.
KASIKORN Business-Technology Group (KBTG) is the IT arm for Kasikornbank (KBank), a large banking group in Thailand with approximately 17 million customers. KBTG delivers a range of technology services to help drive service efficiency and enhance competitiveness.
KBTG’s mission is to help KBank create the digital bank of the future, providing customers with fast, easy, and safe access to financial services from anywhere, and at any time. As KBTG delivers new digital services, its ability to accurately assess each customer’s needs and potential risk is vital to its success.
“In the past, we looked at data to understand what happened last month or last year,” said Tul Roteseree, deputy managing director at KBTG. “We wanted to use machine learning to predict what is going to happen—which loan applicants would likely default or which customers are not only creditworthy, but also credit hungry.”
However, KBTG’s existing data warehouse didn’t provide the scalability or flexibility to effectively support existing Business Intelligence (BI) and analytics, or future predictive analytics needs.
“The more data we can bring together and analyze, the better we can understand our customers,” said Roteseree. “With Netezza, we were limited to roughly two years of pre-aggregated data and not enough capacity to support the full breadth of analytics needed. Our analysts were forced to extract subsets of this data for analysis beyond the canned reports. We needed a modern platform that would allow us to not only consolidate and store more data for longer histories, but also enable more flexible, self-service analytics directly against this shared data.”
KBTG teamed with Cloudera to implement a modern platform for machine learning and analytics that helps improve customer service and fraud detection, and enables innovative services, such as automated lending. With automated lending, the bank can send a targeted loan offer to a customer's mobile app. Once the customer accepts the offer, new funds can be transferred into their account within minutes.
KBTG currently stores and analyzes more than 12 years of transactions in its new platform, and plans to increase that to include nearly 35 years of data. Approximately seven million transactions are added to the platform daily. The ability to include new types of external data, including social media data, will allow the bank to see beyond their own transaction information and gain a more complete view of each customer’s financial needs.
“With Cloudera, we can store a deeper history and a greater variety of data than traditional data warehouse systems, and analyze and operationalize that data all within the same platform so we can be more accurate and understand our customers better and faster,” said Roteseree. “We’re getting approximately 10 times the amount of storage, and 10 times the computing power, and we are well on our way to entirely replacing Netezza.”
As part of its implementation, KBTG is assessing which workloads to run on prem and which are more appropriate for the cloud. “With Cloudera, we have the flexibility to run workloads on-prem, in the cloud or both as needed,” said Roteseree. “We’re exploring moving some of our statement generation capabilities to Cloudera on AWS [Amazon Web Services]. The server that delivers monthly email statements is used quite a bit at the beginning of the month, and then, after that, it sits idle. This type of workload is a perfect fit for the cloud, where we can pay for the processing power as needed. ”
The payoff extends across banking operations and includes the following:
Dramatic reduction in the time to access statement information—from up to a week to seconds. This has yielded a huge improvement in the customer experience and faster response to regulatory or government requests. Additionally, the bank estimates that the ability to provide statements online will save US$5 million in printing and postal costs over five years.
More accurate identification of fraud patterns to reduce the number of false positives analysts must review and better pinpoint actual fraud cases.
Decreased loan approval times and delivery of targeted offers to customers based not only on credit score, but also on interest and need.