Risk modelling is another major use case that’s energized by Hadoop. We think we will find that it closely resembles the fraud detection model use case in which it acts like a model-based discipline. The greater the amount of data volume we have, more we can “connect the dots,” and the more often our output will yield better risk-detection models.
The word “risk” can be used in various contexts and lot of definitions can be drawn from it. For instance, client churn prediction is the risk of a user moving to a competitor; the risk of a loan book impact to the risk of default; risk in health care spans the gamut from outbreak containment to food safety to the possibility of reinfection and much more.
The financial services sector (FSS) is now investing heavily in Hadoop-based risk modelling. This sector seeks to increase the automation and accuracy of its risk assessment and exposure modelling. Hadoop provides the users the opportunity to expand the data volume sets that are utilized in their risk models to include under-utilized sources (or sources that are never utilized), such as electronic-mail, instant messaging, social networking, and interactions with customer service representative and various other data sources. Risk predictive modelling in FSS pop up almost everywhere. They’re used for customer churn prevention, trade manipulation modelling, corporate risk and exposure analytics, and more.
Sometimes an organization issues an insurance policy for natural disasters at home; one problem is clearly identifying how much money is potentially at risk. If the insurer fails to reserve money for the pay-outs, then the regulators will intervene (which the insurer doesn’t wanted ); if the insurer puts more money into its reserves for paying out future policy claims, then they are not able to invest your premium money to make a profit (which insurer doesn’t want too ). We know many organizations that play “blind” to the risk they face since they are not able to run an adequate amount of catastrophic simulations pertaining to variance in wind speed and precipitation rates (among other variables) as these are related to their exposure. Quite easily, these organizations face challenges in stress-testing their risk-predictive models. This ability to manage in more data volumes — for example, weather patterns or the ever-changing socio-economic distribution of their client base — gives them a lot more flexibility and business insight when it comes to create better risk models.
Building and stress-testing risk predictive models is an ideal task for Hadoop. These operations are sometimes computationally expensive and, when we are building a risk model, it is likely impractical to run against a data warehouse