“Data science pays boatload of $$$. I want exactly that boatload of $$$.”
You’ll start searching “Data science tutorial”, “How to do data science?” and Google will throw at you Python, R, unsupervised vs supervised learning etc.
If you are not already a developer, technical analyst of some kind, I would say it’s not a good way how to get to the field.
Before you start, it’s good to realise few things.
- Data is not some mythical beast that needs a magical sword to be conquered. It’s just a footprint of systems and people doing stuff. And in a lot of cases very poor footprint.
- You can learn in incrementally. If you find yourselves in a situation “I have no idea what this is” when trying to learn things. You might be coming to it from the wrong angle.
- Tutorials are not solving any real-world problems. Importing a file with a Python script is not a problem. Learning linear regression and knowing how to plot a regression line in R is not a problem.
Psst, here is a secret – you, most likely, already know what problems you want to solve. Tickets getting stuck somewhere in your CRM process is! Delayed payment is a problem!
A company won’t pay* you for walking around with a drill but surely will appreciate identifying a wall you need to fix, even though you might not know how to fix it.
So where to start?
- Grab a tool of your choice. Find a problem of something you are interested in.
For example, there might be possible improvements in your case management system, since you have got a lot of complaints. - Think about what the dataset represents, make sure you understand what the limitations and issues are. Most likely, the dataset does not describe reality perfectly.
- Think about what you want to measure. Start with simple counts, sums or averages. Break it down by different fields like status or type.
- Plot data into bar charts, timelines, basic histograms and just mess around. See if you find anything interesting. This a part where knowing business and processes is your massive advantage.
- Try to create some groupings, if something pops up. Enrich the dataset with something else. Enrich = VLOOKUP in a lot of cases.
- Just keep iterating. Try different things to learn to approach problem incrementally.
- Find someone you can bounce ideas off. It is likely you already have some people in your company who will be happy to help you.
- Don’t let yourself down, if you won’t be able to find anything revolutionary. Even though industry sells the idea that there always some gold nugget, more often there is not one.
*Truth is that some companies will. The question is if you want to be lucky/unfortunate to be in that situation.
Just above average 