Surely we can train computers to solve some of our most pressing problems?
Data science is leading a revolution in how companies are innovating and competing.
It is being enabled by the vast quantities of data that are being generated as our world population interacts with the internet, and whole economies transition to being ever more digital and connected.
What is not happening organically during this digital transition, per se, is a focus on data collection. It means that most companies are not collecting the right data.
If we are to attempt to train computers to help us to optimise our world, we need to teach them how to make good decisions, and doing that requires us to create good examples we can feed into the training process.
Providing computers with the right inputs, and the desired outcome, is the requirement; it creates what we call Training Data. With amazing training data, we can teach computers to do amazing things.
Getting started means firstly identifying the amazing things we want the computers to do (data exploitation strategy), then secondly to figure out how to collect the data that would train them to do it (data collection strategy). The actual training of the computers is done during the Data Science work (Hypothesis Testing) where we feed the example training data into the machine learning algorithms.
In many ways what data science really discovers is the impedance between the training data needed to achieve the results we want, against the quality of the training data that has been collected.
ByteSumo have developed a methodology to help companies to plan, prioritise, and enable their data science hypotheses.
It means we can de-risk your investments in data science.
To support our view that data science and machine learning is going to have a huge impact on our future, we sponsored S2DS.org, a data science school that ran in September 2014. The school helped train 87 analytic PHD grads into commercial data scientists.
For more details, visit their website, or see our announcement.