Issue 5: Pick a data collection setup that suits your organization’s capabilities and budget. — Cliché: Without data, there can be no data science. But it is true. While learning data science, we mostly use public data sets or scrape data off the web. But in ML-assisted products, most of the data is generated and collected through business applications. The first step in any data…

Issue 4: Perfect is the enemy of good. — The most frequent question I get from developers is: what is the best way to get into Machine Learning? A few years back, my response was: Google for best resources and learn Find problems at work and apply what you learn Repeat Though that response was honest and sincere, I…

Issue 3: Should data science embrace Agile? — To be agile, or not to be, that is the question. I guess the answer depends on whom do you ask. I have seen many Data Scientists bitterly oppose Agile and Scrum: 6 Reasons why I think Agile Data Science does not work Why Scrum is awful for data science Why data science doesn’t respond well to Agile methodologies Don’t make…

Issue 2: Model Evaluation vs. Model Testing vs. Model Explainability. — Is the model’s performance score enough? If your machine learning model has a high correctness score on the holdout test data set, is it safe to deploy it in production? All models are wrong, but some are useful. — George E. P. Box (famous British statistician) But the question I am asking is: Are more correct…

Issue 1: ML products require 5 disciplines — product, data, ML, dev, ops. — When we look at Machine Learning through Software Developer’s lens, building a successful ML product or, more often, ML-assisted product features takes 5 disciplines to come together: product, data, ML, dev, and ops. Product Design & Management encompasses the whole gamut of things: identifying the user and business needs, designing…

Systematic way of collecting data and extracting actionable insights using the Drivetrain Approach. — “Let’s collect all data we can, and we will fish for insights later.” Have you heard this before? That approach seldom works. On rare occasions when it does work a little, the RoI is very low w.r.t. the cost of collecting, processing, and storing…