A year ago I was an interested bystander to the field of Data Science and Machine Learning. From that perspective it seemed that there was a lot of magic involved - the high priests (mathematicians) with pages full of strange symbols and black box code that proclaimed "trust me for the answers".
With a long experience of coding but without even a Maths A-level to my name neither approach seemed right. I decided that I was going to be an interested bystander no more.
Over the last year I have been evaluating machine learning and comparing it with more familiar approaches, not just on canned data sets from the machine learning community, but also for real production applications.
It's meant learning some maths - but not as much as I had feared - and a whole load of new terminology. But I have learnt that you can teach an old programming dog some new machine learning tricks.