A lire sur: http://www.atelier.net/node/404104
In a world of Big Data, many firms are turning to
predictive analytics. However, data analytics cannot be fully
substituted to a human expertise.
Interview with Tapan Patel, Global Marketing Manager at SAS at the Predictive Analytics World Summit.
L’Atelier : How can data inform decision-making in an organization ?
Tapan Patel : The subject of the year is Big Data: the growth in the volume, the variety and the velocity of data. Therefore, the first question is how can predictive analytics companies leverage big data; in terms of data management, getting the data ready for analytics purposes and applying different analytical techniques to solve variety of business problems. Second, how the insights you get from analytics have to be integrated into business processes to be really effective. The question is how organizations must build a culture of analytics among employees and across business units. They should be constantly sure that their employees have the training and the skill sets to leverage data analytics, and help them take the right decision in real time, thanks to this data.
L’Atelier : What is the advantage of relying on the services of an external data analytics organization rather than opting for a home made solution inside your company?
Tapan Patel : A home made solution can’t scale when it comes to applying analytics quickly for a variety of problems. For the company, it would mean having to develop every model from scratch, and this takes time. If you want to take advantage of moving quickly in the market, you should apply analytics fast to take fact based decisions. On the contrary, a commercial solution can help customers jump on and use variety of data sources, quickly apply new kind of pre-built modeling techniques, and automate model comparison and validation steps. However, this kind of solution is not 100% adapted to one’s firm, so that is why the software provided has to be complemented with best practices and domain expertise to fully meet the needs of the customers.
L’Atelier : To what extent will data-driven analytics make reliance on conjecture a thing of the past?
Tapan Patel : We have to make adjustments. In many analytics projects, data miners and statisticians take 70% to 80% of their time to data preparation. Data management, cleaning up the data, and treating them to make them ready for analyses is very critical for any kind of analytical project. The question is to what extent you can trust all of the data, given that there is necessarily some noise among them, and how far they are relevant for the problem you are trying to solve.
L’Atelier: So if you can’t trust data fully, is means human expertise still has a role to play in data analytics?
Tapan Patel:Social media data are especially hard to leverage. How can you assign quality to a tweet? How can you find who is relevant on social networks? In fact, predictive analytics is a science that will help you get from point A to point B, but the art comes in when you have to rely on the business expertise of the person who is going to take the decision in the end, and who has to take into account the weight of the noise. Finally, the best analytic insight comes out from a combination of an art and a science. Then, the analytic insight has to be quickly put in the hands of decision makers or integrated into business processes to take best decisions.
L’Atelier : How can data inform decision-making in an organization ?
Tapan Patel : The subject of the year is Big Data: the growth in the volume, the variety and the velocity of data. Therefore, the first question is how can predictive analytics companies leverage big data; in terms of data management, getting the data ready for analytics purposes and applying different analytical techniques to solve variety of business problems. Second, how the insights you get from analytics have to be integrated into business processes to be really effective. The question is how organizations must build a culture of analytics among employees and across business units. They should be constantly sure that their employees have the training and the skill sets to leverage data analytics, and help them take the right decision in real time, thanks to this data.
L’Atelier : What is the advantage of relying on the services of an external data analytics organization rather than opting for a home made solution inside your company?
Tapan Patel : A home made solution can’t scale when it comes to applying analytics quickly for a variety of problems. For the company, it would mean having to develop every model from scratch, and this takes time. If you want to take advantage of moving quickly in the market, you should apply analytics fast to take fact based decisions. On the contrary, a commercial solution can help customers jump on and use variety of data sources, quickly apply new kind of pre-built modeling techniques, and automate model comparison and validation steps. However, this kind of solution is not 100% adapted to one’s firm, so that is why the software provided has to be complemented with best practices and domain expertise to fully meet the needs of the customers.
L’Atelier : To what extent will data-driven analytics make reliance on conjecture a thing of the past?
Tapan Patel : We have to make adjustments. In many analytics projects, data miners and statisticians take 70% to 80% of their time to data preparation. Data management, cleaning up the data, and treating them to make them ready for analyses is very critical for any kind of analytical project. The question is to what extent you can trust all of the data, given that there is necessarily some noise among them, and how far they are relevant for the problem you are trying to solve.
L’Atelier: So if you can’t trust data fully, is means human expertise still has a role to play in data analytics?
Tapan Patel:Social media data are especially hard to leverage. How can you assign quality to a tweet? How can you find who is relevant on social networks? In fact, predictive analytics is a science that will help you get from point A to point B, but the art comes in when you have to rely on the business expertise of the person who is going to take the decision in the end, and who has to take into account the weight of the noise. Finally, the best analytic insight comes out from a combination of an art and a science. Then, the analytic insight has to be quickly put in the hands of decision makers or integrated into business processes to take best decisions.
Aucun commentaire:
Enregistrer un commentaire