Data Scientists in Banks
Well this is more like why do you need to hire a data scientist for every bank. As of now with the emergence of Big Data Tech many banks have adhered to this concept of hiring data science grads and its more like a trend in countries like Singapore. In countries like Sri Lanka, its definitely an upward tread but there is a lot of catching up.
The objective in hiring a Data Scientist is all about cutting cost and identify new revenue patters. More simply said; its about searching avenues for new money and fine tune the existing processes. To be a Data Scientist in the banking domain, an individual has to be really good with domain knowledge. Given a field like banking, domain knowledge is a key.
If you are curious about money, how its created and valued, the role of a merchant bank when compared to the central bank etc…and got the right tech skills, you should seriously consider your self being a Data Scientist. Capability to dig into data in a way which no one has never done and identify patterns which anyone never imagined of can do wonders to an organisations like banks.
Traditionally the management will have access to reports on daily occurrences of past activities. eg: Transitional Info, Customer Usage , Terminal Usage. Business Intelligence teams will evaluate these and will make predictions for the next quarter. It may work or it may not, which a certain individual cannot guarantee. Also the source for the report may not have all the information captured due to complexions. Eg: most Reports are generated in software systems based on structured date. Which is from data stored in a relational data bases.
(most people misunderstands big data as the increase of data overtime in a RDBMS,Big data technology does consider structured data but its more into tapping into untapped data sources like logs ; dumps; archives)
In other words, only about 10% of data generated from customer activities are converted to structured from and saved in a RDBMS. The rest will be saved as logs or dumps. Usual SQL language which is mostly used for report generation will capture only the data saved in Tables. The rest will not be used or wasted.
How can an data scientist help you with all these? Firstly the number of sources will increase. He or she can will check for the simplest activities reading logs,dumps and also RDBMS. Also the predictions will be done with vast amounts of source information unlike previously and given with a guaranteed accuracy rate. In very simplest form, a data scientist can give you the expected outcome of the next day, week or month with an accuracy of 90% or more.
As a CEO of a bank, what you really need to think is, if your data science team can save you more, than what you pay them or earn you more than what you pay them, then its a win for you.
As per what I have read on the internet most banks who does not want to hire a data science team, thinks that purchasing a BI tool (comes with huge price tags) will solve all these. Machine Learning Systems or BI systems are not like software solutions. Example, you can get your ATM switching software from the same vendor which Bank X uses. But its not easy to plug in a machine learning product since each bank has a different customer bases, different visions and missions, different revenue models etc.
As an example, X Bank has the most customers, which means the number of accounts are high, but does that mean they have the most savings ? No. Y Bank has acquired the most number of ATM transactions, does that mean they have the biggest customer base? No.
So my point is, the behavior of each bank is different from the other.. So does the BI team. Therefore the predictive models has to be custom made… at-least till you get the hang of it and your data science team recommends you to do otherwise.
few areas where a data scientist can do wonders in terms of costs and revenue gain are given below..
- Forecasting ATM cash demand.
- Prevent fraud in Transactions.
- Predictive Maintenance for all Systems. (Including Banking Applications,DB Software and Hardware)
- Predictive Analytics in Customer behavior, personalized banking.
- Target Marketing.
- Identify new Revenue Patterns within Existing Models.
Well these are very high level categories and each and every category can be segregated into 4–5 sub sections. First three categories are mainly for cost-cutting and the last 3 are a mix of cost cutting and gain additional revenue.
I will be hoping to discuss each section in detail later. Leave your thoughts below….
Reference :
https://medium.com/datadriveninvestor/big-data-analytics-in-the-banking-sector-b7cb98d27ed2
https://www.evry.com/globalassets/insight/bank2020/bank-2020---big-data---whitepaper.pdf
https://thefinancialbrand.com/64166/banking-big-data-advanced-analytics-ai/
