Very often analysis is outcome of the questions that we ask and the answers that we seek. When the available data was limited, it was fine to use this technique. We could ask the data to be presented in many different ways by asking questions in different dimensions and data would be churned to give us the answers. But with rapidly growing volume, velocity and variety of data we don’t even know what questions to ask.
Usually we ask questions based on our past experiences. We try to extrapolate the past to predict the future. Our vision and ability to ask question was limited by this finite data set. Now we have so much more data. Not only that, now data that is available is from so many unexpected sources that we really don’t know what to look for in it. With the world gearing up for IoT phenomena almost every device would be a source of data on a continuous basis.
To make sense out of such humongous data, the data scientist are approaching this data in such a way that you let data itself ask the questions. Think about it once again, instead of we framing the questions, now we let data understand itself and frame right questions and answers to that enable us to take actions. Though it is early days, we already have glimpses of such systems being in place such as IBM’s project Watson. As this technique matures and becomes easily available, we would be moving towards commoditisation of self learning systems.
Imagine the world where our mobile devices, health gadgets, our home appliances, text messages, emails, all mobile apps data, all locational data are being analysed 24×7, the systems would know so much about us that it would start predicting our behaviour. We have already seen the retail companies already knowing our buying behaviour based on our our buying pattern. Now with this kind of information available the system would churn out its own hypothesis and may would tell publishers what kind of books should be written.
We can imagine similar impact on the industrial behaviour. Todays systems are capable of predicting maintenance time and parts replacement. But going forward with the help of data from every possible sensor, system would tell us what to manufacture and what should be the features.
Interesting times are ahead.