TIPS:
β Start with the simplest models
β Identify the patterns from data
β Let the data speak
β Don't force data to cry
β Spend most of your time in EDA, preprocessing and feature engineering
β Choose the tech stack with which you are more familiar
β Try to involve people with domain expertise.
β Visualize your data. It helps a lot in understanding the patterns.
β Use statistical inference to understand the underlying distributions
β Initially solve what is possible
β Do market research
β See what other data scientists are doing in the same domain
β Try to relate the results of your work with the business value
β Communicate your findings
β Want to add anything?
Feel free to share your thoughts in the comments below!
Top comments (6)
Excellent and often overlooked tip. It's easy to get swept away in grand ideas that are very hard to achieve. Start with what's possible and grow solutions from there.
Yes. That's the most overlooked point. People want to start big instead, they should take baby steps to approach the problem.
Consider possible flaws and biases in your data and how that might negatively affect your results.
Thanks for the amazing addition to the list, Ben!
How I want to learn this can teach me
Hi. If you want to learn Machine Learning, I have listed Free ML Courses here: git.io/ai-resources.