You really want to avoid going back to school just to work on projects that will be your portfolio. You are busy. Interviewing is hard. Data Science is new and means different things to different people. And most of all, you're worried that since you are learning skills on your own time at home, you may not be learning the right things. So when you construct your data science portfolio, it may be all wrong.
Your data science portfolio should work for you
Your dream is to have created a portfolio that gets you a job offers without you needing to reach out to employers. You want recruiters and employers reach out to you cold to offer you employment opportunities based on your portfolio. You'd love the portfolio to increase your name recognition in the field you want to move into. You'd like your future employer to already know you by reputation because they've already seen your portfolio.
You stand out against all other people interviewing because of your portfolio
You’d love to ace the interview section by showcasing your portfolio. You want to showcase your skills and stand out against all the other people interviewing for that data science job you want. You really want to prove to the employer that you're worth it even though you are changing careers. You want your portfolio to showcase your critical thinking skills so well that critical thinking questions that don’t come up in interviews because it's extremely clear from your portfolio that you can do critical thinking. You also want your communication skills to never be questioned.
You’ve read about people who flunked the interview but still got the offer after they showed the hiring manager their blog in a post-interview email follow-up. You don't want to flunk the interview. But if it happens, you'd like to still get the offer based on your portfolio.
You are going to build a data science portfolio
The question is - where to start. How can you make sure that it'll look great before, during, and after the interview. How can you make sure that your portfolio works tirelessly for you.
How You Should __NOT__ Create A Data Science Portfolio To Show Employers
The boil the ocean approach to building a data science portfolio. The boiling the ocean approach is when you do random projects in hopes that you learn something of use to a future employer. Because there are so many different definitions of data science, and so many different team make ups, and so many different languages, and so many different techniques, and so many different goals, the chance that you do something useful to a future employer is minimal. Which means you won't stand out and you won't be using your time wisely. You're busy. Shouldn't you be working on things that actually help you?
How You __SHOULD__ Create A Data Science Portfolio To Show Employers
The "magnifying glass fire starting" approach to building a data science portfolio. You can use a magnifying glass to start a fire because it concentrates the sunlight going through it to such an intense degree that the concentration of heat can reach incredibly high temperatures. You want to do the same thing with your portfolio - make it so concentrated and intense that your future employers will salivate at the thought of having you join their team.
Here are the 10 steps to create a data science portfolio that will get you hired:
- Forget about boiling the ocean
- Take a hyper-targeted "magnifying glass fire starting" approach
- Find 5-10 data science jobs you’d take if offered to you
- Figure out what skill sets / job responsibilities you would have
- Find the common ones (NLP, recommendation, classification, etc.)
- Figure out the tool sets that the jobs require
- Find the common tools (R, Python, Scala, Hadoop, etc.)
- If it's not a tool or tools you know, then learn them as part of the portfolio work
- Do 3 projects that cover the common job responsibilities for the jobs you are interested in using the common tools for those jobs
- Do a structured writeup for each of the three projects.
How to think about step 9 "Choosing Projects"
Don't over think it. After having looked at the data of the types of data science jobs you want to do, you should have an okay idea of potential projects. Rather than worrying too much about it, start the first project. Then you can re-evaluate what you will do for the second one. Then after the second project, you can then re-evaluate what you do for the third project. As you can imagine, you'll have clearer and clearer thoughts as you go through these projects from start to finish. You may even surprise yourself by choosing a data set and project that didn't even cross your mind when you came up with the initial list.
How to think about step 10 "Structured writeup of each project"
In addition to doing great work, a key part of the portfolio is being able to showcase your critical thinking and communication skills. This is the step that is the most important in showcasing those two skills. This step will take a good amount of time. When writing up your project, here are some questions to keep in mind. In fact, you might even have these as individual sections. There are 20 main things to cover:
- What question you are looking to answer?
- Why does this question matter?
- What data did you use?
- Where you got the data?
- How was the data sampled?
- How was the data obtained?
- How did you explore the data?
- How did you model the data?
- Why you chose to model it that way?
- What code did you write / use?
- How did you fit the model?
- How did you validated the model?
- How you know the results make sense?
- How did you visualized the results?
- How you would communicate the results to others?
- What did you learn?
- What you would do differently if you did this project again?
- If you were going to continue this work, what next steps you would take with this project?
- How you would explain what you did to a data scientist?
- How you would explain what you did to a non-data scientist
Again, like with step number 9 - the key here is to write something down, not to have some be perfect. After you do this for the first project, you'll have a better sense of how to think about project number 2. When you do the write up of project number 2, you'll do even better because you've done it before and you'll have a better feel for how the project came together. Then when you do project number 3, you'll do even better than project 2. So on and so forth. Remember - the goal is to write something, not have it be perfect.
To that end, it's time to get started with the first step! Go out today and start looking at data science job postings to see if there are any that you like. If you find one or a few, copy the text down and start keeping notes so that you can started on the 10 steps to create a data science portfolio that will get you hired!