I think there is lack of leadership in the big data space who can say "This is how you are going to analyze data in your system and this is what you are going to use". There is a lack of a company like Apple who can say that "This the way smart phones should be and here's my product iPhone".
I am seeing companies all around me which wants to enter the big data and big data analytics market because its a huge untapped market. No clear market leader exists and its a fast growing segment. Companies are recognizing the value of data and performing analytics to gain value from data to gain competitive advantage.
Why are the companies trying to sell analytics solution not able to choose a segment like fraud detection? I am unable to understand why fraud detection agencies have to use ten tools to have an application running? Why is there no single solution for complete analytical solution in market? The solution should do model generation (Run repeated regression analysis with various parameters automatically, run k-mean clustering, run APRIORI and such), model validation and real time data incorporation that can solve all the problems of a fraud detection system. As a fraud detection application architect, I just want to go to one tool which will allow me to automatically generate models (smart) and suck in data what will keep on validating the models on regular basis and define rules for making real time decisions per transaction to determine if it is fraud or genuine.
I am seeing companies all around me which wants to enter the big data and big data analytics market because its a huge untapped market. No clear market leader exists and its a fast growing segment. Companies are recognizing the value of data and performing analytics to gain value from data to gain competitive advantage.
Why are the companies trying to sell analytics solution not able to choose a segment like fraud detection? I am unable to understand why fraud detection agencies have to use ten tools to have an application running? Why is there no single solution for complete analytical solution in market? The solution should do model generation (Run repeated regression analysis with various parameters automatically, run k-mean clustering, run APRIORI and such), model validation and real time data incorporation that can solve all the problems of a fraud detection system. As a fraud detection application architect, I just want to go to one tool which will allow me to automatically generate models (smart) and suck in data what will keep on validating the models on regular basis and define rules for making real time decisions per transaction to determine if it is fraud or genuine.
4 comments:
I blog quitе often and I truly thanκ
you for уour іnformation. Your aгtіclе has truly peaked
my іntеrest. Ι'm going to take a note of your site and keep checking for new details about once a week. I subscribed to your Feed as well.
Also visit my page; http://homeinsurancecomparisonuk.co.uk/home-contents-insurance.html
Wonԁeгful blog yоu havе heгe but
I wаѕ wanting to know if уou
knew οf аny mеssage boаrds that сover the
sаme topiсs tаlked аbout in this artiсle?
Ι'd really love to be a part of group where I can get comments from other knowledgeable people that share the same interest. If you have any recommendations, please let me know. Thanks a lot!
Also visit my site; home contents insurance
Thanks, that is a very interesting challenge to the space. I think streamlining the tools for a fraud detection application could be really valuable. I believe you could pose that same challenge in other areas of analytics applications as well, such as predicting system failure or even in something like customer segmentation.
Of course, in many advanced analytics applications the data preparation tends to be the largest piece of work. Do you see this as being part of the same tool, or do you imagine a tool that applies models (and optimizes and evaluates them) after the data has been put into a standard format for consumption?
@Chance Coble -> I see industries specializing in verticals and providing targeted application which already chooses the exploratory models applicable to specific industry as well as predictive models. Provide default implementation with some basic attributes chosen and list of attributes proven to be redundant. Or an automated way of feeding from exploratory algorithms to predictive algorithms.
Post a Comment