With an ever-increasing variety of options, I found trying to decide how I wanted to further my data science skills challenging. I'd already completed a number of online MOOCs (such as Andrew Ng's famous Machine Learning Coursera course), however knew that I wanted some more formal, in-person tuition. I was/am a Python developer and Data Engineer by profession, and so wanted to find a course that would spend relatively little time on the basics of Python programming and jump straight into the really interesting stuff (note: most of the class had far less programming experience – there’s no requirement to already be very experienced with Python!). I'd already heard of/taken free classes with well-known providers such as General Assembly and DataCamp; it was whilst looking at reviews for their Data Science bootcamp offerings that I came across Cambridge Spark. I was attracted to the breadth of topics in the curriculum (everything from linear and logistic regression to neural networks, natural language processing, and big-data processing with Apache Spark) and the fact that I'd be receiving a full weekend of tuition (9:30 am to 5pm, Saturday and Sunday) from highly experienced practitioners every fortnight for nearly 6 months. I knew I wouldn't be able to take 5-8 weeks off work to complete a full-time bootcamp, and preferred being able to devote weekends to learning, rather than spending a couple of hours in the evening twice/three times a week after work (when I'm already knackered). Additionally, compared to other part-time bootcamps, the Cambridge Spark offering presented a very competitive hourly rate for the amount of in-person tuition received. I took the plunge, and was not disappointed. The quality of the materials and instruction far exceeded my expectations. Time during the taught weekends was usually split roughly equally between informal lecture style tuition and hands-on exercises via Jupyter notebooks; allowing students to gain experience with the concepts being taught, and have all questions answered by the tutors. The weekends were held in a perfectly pleasant function room at a hotel near Kings Cross; however future students will benefit from the custom teaching facilities that Cambridge Spark have recently opened (again near Kings Cross). Between the taught weekends there was additional recommended reading, the ability to re-listen to recordings of the taught materials, and assignments to complete via Cambridge Spark's proprietary learning platform KATE (a real jewel in their crown). At the end of the course I spent 6 weeks on a capstone project – working with a start-up on a real-life problem to develop object detection models. Whilst the high-level theory behind the various machine learning algorithms covered is very well taught, the course primarily focuses on practical implementation. Those that complete the bootcamp will be well versed in how to go about approaching data science problems, including: data cleansing and feature selection; using Numpy, Pandas and scikit-learn; selecting and calculating appropriate performance metrics for models; performing hyperparameter tuning; and having the knowledge to justify their choices and be able to explain how their models work. Personally, I’m now very keen to gain much deeper knowledge of the underlying mathematics of machine learning algorithms, such that I may be able to contribute to the development of new techniques – hence I will soon be starting a full-time MSc in data science. However, I am very glad to have first completed the Cambridge Spark course, which has provided all of the skills I need to move from data engineering into data science. I honestly can’t recommend Cambridge Spark highly enough. (P.S. you also get to attend a formal dinner held at a Cambridge University college; what other bootcamp offers that?!) | 