KDnuggets Poll: Future of Predictive Analytics: Human or Machine?
The robots are taking over many jobs - will they take yours and mine? New KDnuggets Poll is asking if and when automation will reach the level of human data scientists.
By Gregory Piatetsky,
@kdnuggets.
The robots are taking over many jobs - will they take yours and mine?
Will Data Scientists be replaced by software, and if so, when?
Please vote in New KDnuggets Poll below.
This poll is closed - here are the results
Data Scientists Automated and Unemployed by 2025.
Here are some of my thoughts on the topic.
Justin Washtell argues in The Imminent Future of Predictive Modeling that the future of Predictive Modeling is automation.
Clearly, there are not enough Ph.D. Data Scientists for all analytics tasks and "Democratization" of analytics - making analytics easier to use by non-data scientists - is a big trend. This trend requires automating many parts of the data mining process, like feature selection and (some) data cleaning/preprocessing. Taken to its logical conclusion, "democratization" will inevitably lead towards "automation" of analytics.
Automatic Statistician, DataRobot, Google Analytics (with its Intelligence alerts), IBM Watson Analytics, KXEN Modeling Factory (now part of SAP), and other companies and projects are already automating many predictive analytics tasks.
On the other hand, data scientists like Mikio Braun warn that Data analysis is hard and "you can't just give a few coders new tools and they will produce something which works. "
I think there is no contradiction between these points of view:
because there is a "frontier" of automation, as shown below. How easy it is to automate Data Science tasks depends on how complex the tasks are, amount of domain knowledge required, etc.
Some tasks - like online ads marketplace - are already automated, with gazillions of ad-related predictions and decisions done fully automatically every day.
Some tasks are partly automated, like fraud-detection, medical image analysis, and options trading (read about a twitter bot that made millions on the options market).
However, the frontier of automation is moving and getting closer to human level. Analytics guru Tom Davenport writes that marketing activities will also become increasingly automated and
Data Scientists who want to keep their jobs in the future will also need to be very creative and have not just technical, but human-oriented communication and presentation skills.
Please vote in the KDnuggets Poll above.
Related:
The robots are taking over many jobs - will they take yours and mine?
Will Data Scientists be replaced by software, and if so, when?
This poll is closed - here are the results
Data Scientists Automated and Unemployed by 2025.
Here are some of my thoughts on the topic.
Justin Washtell argues in The Imminent Future of Predictive Modeling that the future of Predictive Modeling is automation.
Clearly, there are not enough Ph.D. Data Scientists for all analytics tasks and "Democratization" of analytics - making analytics easier to use by non-data scientists - is a big trend. This trend requires automating many parts of the data mining process, like feature selection and (some) data cleaning/preprocessing. Taken to its logical conclusion, "democratization" will inevitably lead towards "automation" of analytics.
Automatic Statistician, DataRobot, Google Analytics (with its Intelligence alerts), IBM Watson Analytics, KXEN Modeling Factory (now part of SAP), and other companies and projects are already automating many predictive analytics tasks.
On the other hand, data scientists like Mikio Braun warn that Data analysis is hard and "you can't just give a few coders new tools and they will produce something which works. "
I think there is no contradiction between these points of view:
- Predictive Analytics needs to be automated
- Predictive Analytics is hard
because there is a "frontier" of automation, as shown below. How easy it is to automate Data Science tasks depends on how complex the tasks are, amount of domain knowledge required, etc.
Some tasks - like online ads marketplace - are already automated, with gazillions of ad-related predictions and decisions done fully automatically every day.
Some tasks are partly automated, like fraud-detection, medical image analysis, and options trading (read about a twitter bot that made millions on the options market).
However, the frontier of automation is moving and getting closer to human level. Analytics guru Tom Davenport writes that marketing activities will also become increasingly automated and
Those who keep their jobs will have found a way to augment the work of smart marketing machines. After all, marketers are supposed to be creative!
Data Scientists who want to keep their jobs in the future will also need to be very creative and have not just technical, but human-oriented communication and presentation skills.
Please vote in the KDnuggets Poll above.
Related:
