Hi Yuliya!
This is a tough question, and you could write several books without answering each part fully -- I'll do my best to give a quick overview. I think the best way to gain this experience is to do it over and over.
Usually, the question you're trying to answer is a business one, and the process goes like this (I'm paraphrasing a slide by Renee Teate here):
Business question --> data question --> data answer --> business answer
So as far as how to frame the analysis, I start by thinking about where I want to end up -- like what it is I actually want the analysis to tell me, and work backwards from there. Is it a prediction? Is it a summarization of historical data? Is it an analysis, a model, a service? Asking these kinds of questions helps me to frame how I should approach my problem, what data I should use, and what I should be working towards. Asking these questions up-front is also helpful so that you know what you *can't* accomplish. If you realize that you don't have the data to actually help with the problem you're trying to solve, you can save a lot of time by re-phrasing the question or further communicating with stakeholders to refine the question and approach(es) you could take.
For examples, I'd suggest looking at some of the kernels on Kaggle that are focused on how a person approached a problem. It's really hard to find examples of a fully described problem space, the attempts a person made (and threw out) to solve the problem, and how they made those decisions -- although these do make for good conference talks, so it might be worth Googling those.
Hope that helps!
Caitlin
------------------------------
Caitlin Hudon
------------------------------
Original Message:
Sent: 10-05-2018 13:02
From: Yuliya Astapova
Subject: Ask a Data Scientist: Careers in Data Science - LIVE SESSION (Oct 4, 2018)
There are lots of resources online for learning data science skills like Python, R, or machine learning, but there are not as many that talk about how to actually set up and solve data science problems. How do you define a question you want to answer, and how do you design your project or solution? Are there any resources you recommend to gain more experience in this?
------------------------------
Yuliya Astapova
Original Message:
Sent: 09-21-2018 02:26
From: Gabriela de Queiroz
Subject: Ask a Data Scientist: Careers in Data Science - LIVE SESSION (Oct 4, 2018)
Hello everyone,
I'm excited to announce the "Ask a Data Scientist: Careers in Data Science" - LIVE EVENT!
Join us for this exclusive community event on Oct 4, from 11:00 AM to 12:00 PM (ET) where I will have a pleasure to chat with the great Data Scientist Caitlin Hudon live via webcast. We will discuss topics such as breaking into and transitioning to data science, some of the skills required, how to prepare for job interviews, and much more.
We will be taking questions during the session as well as answering the questions you leave here in this forum. Feel free to drop them ahead of time and we will make sure to answer them.
We look forward to having you at this special community event!
Cheers,
Gabriela de Queiroz
Register here: http://ibm.biz/ask-a-data-scientist
------------------------------
Gabriela de Queiroz
------------------------------
#GlobalAIandDataScience
#GlobalDataScience