Using the past to tell the future: how to rely on Big Data [financeME]
(financeME Via Acquire Media NewsEdge) If I asked you to close your eyes - and then, with them still tightly shut, tell me what you were going to see when you opened them, you could only guess. Your best bet would be to presume everything would be the same as before. Yet, anything could have happened during that short period.
In fact, nobody in business and especially in banking over the past two or three years could say that life - or the markets - are wholly predictable. But we still rely heavily on past patterns to help us anticipate future events and behaviour.
Most of the analytical techniques we use to create business intelligence in the financial services industry involve looking back at what has happened and modelling the future using this information. Data mining, for example, is a highly valuable activity, but it is based on what has gone before, and post-financial crisis the world is a different place from the past and outcomes are not so certain.
This does not detract from the value of data; rather, it indicates that we should be smarter about the way we use it. It is, perhaps, the one asset which is worth more post-crisis than before. During a particularly tense week in the early stages of the crisis, one bank reported that data queries went up by 600 per cent as senior executives, shareholders and politicians alike all asked pertinent questions.
However banks and other financial institutions with only operational IT systems or where data is kept only in department-specific silos are at a disadvantage. Operational systems have to be very good at doing what they do, to be very good at doing what they do, such as running current accounts. They are honed to answer the questions we know they are going to be asked. But, when it comes to the business as a whole, nobody knows what the future questions might be and this requires a different technological approach.
If I gave you a pack of cards and asked you to give me the three of hearts, you would need to take time to look through the pack to find it. An operational system knows I am going to ask for the three of hearts, and so it has it on top of the pack, ready. However, what happens if I randomly ask for another card; the ace of spades, for instance? An operational system is not designed to respond to random questions like this, however a data warehouse is - and it adopts a specific approach to do so. A data warehouse does the equivalent of giving 52 people each one card so that someone comes up with the requsted card immediately.
This is how massive parallel processing in an enterprise data warehouse works. Our analytical tools are not primed for any standard interrogations, because there aren't any.
Instead it is configured to respond as quickly as possible to the unexpected.
This approach enables repositories of broadly-based, pan-organisational, integrated data that can help address complex issues such as risk, regulatory compliance, financial management - and even to help shape future product or service development.
The real art is in using this rich mix of data to help foresee events as they are about to happen. For example, data taken from a website may be used to alert a call centre that when a customer contacts them, they will be interested in, say, mortgages or insurance, as these were the products over which their computer mouse was hovering. If this "event data" can be combined with information about the profitability of this particular customer and their propensity to buy, then the information is worth even more.
The same is true of assessing credit risk. Banks can pre-empt a problem by combining up to the minute event data such as spending profile, number of trips to get cash from an ATM, and visits to an online account with analytical data from different sources such as credit rating, total amount of credit accumulated.
= The conjunction of profitability with risk data is also particularly important post-credit crunch. Using the right integration of information, banks can enhance their management of capital, ensuring the best return and sandbagging against unexpected losses in a volatile environment.
But one of the main reasons why data warehousing is more important now than ever, is because it is an investment in future change. Once a data warehouse is in place, there is no limit to the amount of event management, calculations, analysis, reporting or decision making it supports. Consequently, what can appear to be a major challenge for one bank, is a small issue that can be solved almost instantaneously for another with the help of a data warehouse.
For example, one European bank I know had to implement software to deal with an important regulatory requirement. It didn't have an enterprise data warehouse and ran out of time to implement one. It had to resort to an individual solution instead, even though this involved coding tens of thousands of lines of data. A few years later, further regulations came into force and this prompted another serious upheaval for it. Yet, another similar bank that had implemented a data warehouse at the beginning of the process, hardly noticed the more recent regulations as compliance had become a straightforward issue.
There is little doubt that we are currently experiencing an era of consolidation - not only between different banks and other institutions, but also between departments and functions. The amalgamation and integration of data within the data warehouse is a bi-product of this. For example, the global regulatory standard Basel II, and the emerging Basel III, often involves the financial department doing the reporting, but the risk experts providing the data that supports the calculations needed.
Doing these calculations within the data warehouse where all information is rapidly accessible makes perfect sense. It stores not only the results but also the input data and the calculations themselves to give complete transparency surrounding the concluding figures and the steps taken to reach them.
Data is an expanding commodity. The volume available has risen significantly with developments such as the increasing use of internet banking applications and looks set to continue to grow with our current interest in social media and networking. But banks and other businesses need to make this data work harder for them.
There used to be a misconception that enterprise data warehousing was, in itself, a risky business. But with the experience, knowledge and power that has been developed over the past years, specialists such as Teradata have eradicated any uncertainties, if, indeed, they really did exist in the first place.
The risk in the future is not realising what is possible; and, more specifically, in being unable to reach an answer before your competitors. The winners will be those who prepare a route to those answers; even though they have no idea yet what the questions might be.
Valérie Lourme, Senior Industry Consultant Financial Services, Teradata
Highly-skilled Teradata Senior Industry Consultant. Valérie Lourme provides expert consultancy and guidance to leading financial services companies that benefits from extensive local knowledge and experience of working in retail banking. Prior to joining Teradata Valérie worked on a major compliance project at BNP Paribas, where she also held senior roles in the leading French bank's retail banking division.
Valérie is based in Magny les Hameaux, France. Her specialisations include retail banking, risk management and Basel compliance and insurance.
There used to be a misconception that enterprise data warehousing was, in itself, a risky business. But with the experience, knowledge and power that has been developed over the past years, specialists such as Teradata have eradicated any uncertainties, if, indeed, they really did exist in the first place?
The real art is in using this rich mix of data to help foresee events as they are about to happen?
This does not detract from the value of data; rather, it indicates that we should be smarter about the way we use it. It is, perhaps, the one asset which is worth more post- crisis than before?
(c) 2013 CPI Financial. All rights reserved. Provided by Syndigate.info an Albawaba.com company
[ Back To Technology News's Homepage ]