It is surprising how Technology is making us mere hoarders and collectors of Data rather than analysts of Data! How, then, are we different from those who cram information for the very purpose of spewing it out in an exam or interview only to forget it in a week's time? Our obsession with the term Data as in Data Points or Data Mining has robbed us of the ability to go beyond what the Data means!
Research Studies show that Data Analysts spend 80 per cent of their energy, resources and time Mining Data, a euphemism for hoarding Data and a mere 20 per cent is spent on creating Models on the basis of an analysis of Data. The biggest challenge today is not just how we process data but rather what we do with the analysis of Data. Instead of really working on projections and models based on analysed data, we simply get bogged down by terms like Deep Learning or Machine learning instead of trying to really get to terms with the Data we have collected! Data that is raw and unprocessed is simply random and lacks patterns and trends!
Projecting Data crudely on graphs, pie-charts and bar-diagrams does not constitute analysis, rather it is simply a pictorial form of presenting Data. Unfortunately, many of us are satisfied with this! What matters most is what we do with Data Statistics. The analysis of Data can only come after the Data has been presented and then what follows is the creation of a Model that will attempt to iron out or remove the weaknesses existing in the previous Model.
At the school level, simply collecting Data on poor performers without analysis and Projected trends will not help. The Data needs to be analysed and the analysis should lead to the creation of a model that addresses a specific learning disability. In cases where the Data is simply overwhelming for human minds, there is a need for Artificial Intelligence that can not only analyse Data but also to create models and make projections based on existing patterns of learning. Machine Learning as such refers to exactly this! Take for example Data Mining in Marketing, a Cell Phone company would collect large amounts of Data based on Consumer expectations for smartphones through online surveys on social networking sites so that it could analyse the Data and then make future predictions about what features would be more attractive to consumers. These features could be added in the forthcoming cell phone models to be launched thus making them attractive to customers.
Mobile phone manufacturers make use of Deep Learning strategies in order to predict words that are being typed before the user has typed them, often creating embarrassing mistakes. Alexa, a virtual assistant competing with Siri can listen to your voice commands and then get the job done, whether it is switching on the air conditioning or perhaps even giving you the scores of the cricket match that was played the previous day. All this is possible because the software is able to make predictions based on large amounts of Data that have been analysed and modelled regarding appropriate responses.
Whether we humans might be capable of going beyond Data Mining, without the help of Machine Learning or even Deep Learning is doubtful because of an overwhelming amount of Data. At the school level, it should suffice being able to collect Data Points that are limited to specific problems faced by students and not go too deep into an area that can be handled only through AI. It would great if one ask a virtual personal assistant to create a tailor-made learning module for Tom who has problems with clauses in English, or perhaps Martha who has issues with word problems in Mathematics. However, it is doubtful whether Data Points provided by the sample of students weak in clauses or word problems might really be enough for Deep Learning to provide our virtual assistants with enough insight to create a remedial lesson module.
Whether or not individual schools would have the wherewithal or the resources to provide for Machine Learning or Deep Learning or even enough Data Points based on their limited samples is doubtful. To make effective predictions, models and corrective instructional modules would require Data from all the schools in the country if not the whole world! The larger the Data, the more accurate the projections and models. It would, however, be very wrong to compare students learning in schools to consumers of Smartphones, or consumers of Laptops because students are ever changing, ever evolving and they have more fluid intelligence than adults with crystallised intelligence!