I am preparing to transition from analyst to data scientist role. What are the necessary skills I should learn? A/B Test? Experiment design? Causal inference? I hate that I work in HR (people analytics) and was not able to do a lot of high impact work. I currently use sql, tableau. Been working on python skills outside work (dsa, data manipulation, ML projects) Tc: 70k Yoe: 3
forecasting
Yep, time series analysis (if you meant the one) will be a good pick
Ok, on it. What other techniques you use in most of your projects? Is causal inference used often?
I didn't include hypothesis testing specifically since you mentioned inference already - and these two kinda tied together, but it would certainly come in handy. Overall, the skill set/techniques one will be using varies greatly depending on the industry, company, team, project, etc. So it's hard to provide a one-fits-all answer. What would be your scope of work, OP?
Ok makes sense. Well idk what my scope of work would be. I am just blindly looking to get a DS role in any industry and any company as any position would be better than the one I am currently in. I am just trying to learn all the DS techniques. Do you recommend I filter my search down to specific industry and company and then prepare?
I'd say that this approach would be more effective than basically spray'n'pray-ing, especially given that you can ask around here on Blind (or check Glassdoor's interview reviews) about which techniques DS mostly use in a particular company or industry. Before I start working in Amazon, I had worked for banks and one market research firm and should say that decision trees, regression and KNN were by far the most popular choices at the time.
More Python and SQL coding, theoretical concepts of the following: t-test, p-value, power concepts, sample size estimation, multiple testing, ab testing, least square regression, interaction in OLS, logistic regression, ridge, lasso, t-tests, anova, bayes inference, different distributions (cdf, pdf, pmf), survival analysis, principal components, clustering, time series, decision tree, random forest, xgboost
This is why we can't have nice things😃 Let me guess - you came from tech (i.e., not business) background, didn't you?
Data Scientist @ Visa. Before Visa, data Scientist in two unknown companies. Phd in a health field
Probability, Statistics, Product Sense/Business Intuition. And then apply all these techniques that are useful to solve an ambiguous problem.
I assumed Data Scientists at Meta do modelling. This is product analytics
Yeah but OP is already doing ML on the side. These things are also helpful for DS because they need these skills to identify opportunities
regression, classification, optimization, forecasting and a bit of NLP. causal inference, ab testing is good but not used as much as the former.
Thank you! How about learning azure or getting azure data scientist cert? Will that help to get better applied data science knowledge?
What you laid out are the basics of product analytics, not DS. As someone previously mentioned, focus on regression, ML basics, stat basics and keep your sql and visualization strong.
Thank you for clearing it out. I have been so confused with all these companies asking million different things
For non ML roles: 1. SQL (or better PySpark) 2. Intro to ML (follow any good course) 3. Intro to Probability and Statistics For ML roles: 1. Replace SQL with Data Structure and Algorithms 2, 3. Remains same 4. TensorFlow or PyTorch These can easily get you a DS job. P.S. For senior roles, you need design experience as well
Thank you! I have good understanding of all the non-ML role skills. How can i stand out and get a job? Cuz I am not able to get any interviews and I don’t do a lot of data science at work. Does creating a portfolio help?
Getting the first DS role will be tough and everything after that will be easy. So be persistent! For that first job, you need to try a bunch of things like building GitHub profile, doing projects, writing blogs etc. There is no definite way but if you keep on trying, success will be just a matter of time. Keep your fundamentals strong and keep interviewing.
Sql, Python Frameworks, Probability, Product Analysis Dsa also if Targeting Faang
Thats product analyst man. Its crazy how the data roles are fucked up
SO fucked up! 😪
Try triplebyte; if they have tests for data scientists you'll be in good hands.
World Conflicts
14h
543
Israeli precision-guided munition likely killed group of children playing foosball in Gaza, weapons experts say
Personal Finance
9h
1008
Thank you AAPL and NVDA
Cars
18h
461
Best outdoorsy SUV?
India
15h
612
'Hindutva': The Radical Hindu Ideology That Seeks to 'Push Christianity Out of India’
World Conflicts
12h
402
Why I Find Free Palestine Inspiring
I'd say that statistical techniques (mostly regression of various sorts, factor analysis, etc.) and understanding of ML basics, like clusterization and classification approaches, would be enough
Thank you. What other statistical techniques? Maybe a good understanding of hypothesis testing and?