In this post, I will give a brief summary of my reflection paper. The highlights below are more personal, whereas the completed assignment uploaded to UTSonline contains the artefacts and applications of my skills development.
Challenge: Introduction to Data Mining
I am not grounded in a computer science or math background; I know I will need to do extra work in regards to my learning within this degree. As remarked in my blog posted 19 April 2016, ‘I’m pretty old school when it comes to learning and studying, whereas I still like to take hand written notes as I find it sinks in a little better; I still read, summarise and rewrite those notes; and I tend to go looking for further explanations when something isn’t clicking (and usually text with pictures – as I’m a visual learner). It’s really the simple method of ‘rinse and repeat’ until it sticks. The more I read, the more connections I made. When I layout out information on paper, it forces me to make sense of it, include what is relevant; and add paragraph breaks, bold, italics or images as I see fit’.
I further commented within the same post on the important elements of this experience as follows, ‘to not solely rely on what’s given to you. I think it would be foolish to assume just because we all read the same book, that we would all interpret it in the same way. Everyone is different; I like to digest my learning in simplified chunks, preferably with images or diagrams. I also enjoy reading from different perspectives or angles until I find one that “grammatically paints” a nicer picture of understanding for me’.
Challenge: CRISP-DM Methodology
It’s good to review past knowledge, read other opinions, look at other models and see what would be most practical for your toolbox. I liked the concept of the Data Science Process, and the trigger questions at every stage. My observation of the Data Science Process; it is a simplified version of the CRISP-DM methodology with more prompts along the way, which is helpful as a data practitioner when the data mining process hits a roadblock.
This is not to say I dismiss the CRISP-DM, as posted on 19 April 2016, ‘I like the idea of the CRISP-DM model, although find it overcompensating for some short comings; it would appear that industry is taking the bulk of it and omitting/joining bits here and there – but generally speaking, the concept is still worth its weight in gold’.
My theory as posted 19 April 2016, ‘is that eventually your Google search will pull up a site that’s written in your personally digestible language and things will start to make a little more sense than they did the day before’. This is definitely the case and I am glad I kept searching to find the Coursera course. I know my learning KNIME will most likely be slower than most, but I am not deflated by that.
I’ve also found that reading my peers’ blog posts on CICaround has aided in my KNIME discoveries. Their reflections on data cleaning have proved great insight to the different approaches one can take within the software. And finally, ‘working within a group in this first instance was beneficial as we all have enough knowledge together to pull through to the end. KNIME is a necessary tool within the toolkit, and although I am still trying to figure out how to turn it on (or even plug it in), eventually it will become my hammer’ – as posted in Double Dam: Challenge + Block 1. Yet another gentle reminder of data science being a team sport; not only in practice, but also in learning.
Challenge: Modeling Challenges and Impacts
It was necessary to go elsewhere in order to understand and be able to pass on my learnings to my team in a way that I could confidently explain. I know that whatever is on UTSonline is more of a slice than the whole pie. I went through three iterations of the notes in order to simplify the learning as much as possible. Iteration is definitely a skill all data practitioners should get used to; not only in research but as much in process of discovery as well.
I can’t stress enough of the learning I am getting from the Coursera course and how it has reduced my hesitation to be able to complete this course and further my studies in machine learning. I didn’t enter this course with most of the prerequisites and I knew it wouldn’t be easy, but it doesn’t need to be hard either. I also don’t want to pigeon hole myself and not be able to process data like the data scientist beside me. I am realistic in that machine learning won’t necessary be my forte; but I am also cognisant that it is a huge part of my toolkit and I both need and want to learn more.
- Complex Systems Thinking
- Create Value in Problem Solving and Inquiry
- Creative Analytical and Rigorous Sense Making
- Persuasive and Robust Communication