Welcome Data Crunch Listeners!

3 Tips for Effective Data Science Processes

1. Don't over plan One of the most common data initiative failures revolves around too much central planning. Businesses move quickly, and the questions data scientists are trying to answer move even faster. Any process that’s counting on the world being the same tomorrow as it was today is going to fail. The best teams are often those that ship quickly, see what works, and iterate.
2. There’s no one process that fits everyone. Analysts and data scientists come from all sorts of backgrounds. Some are former bankers and consultants who are most comfortable making presentations to execs in Powerpoint. Some are PhDs who’ve written popular R packages but have little industry experience. Others have been on frontend engineering teams before getting deep into data visualization. And others still are operations specialists who wrote managed enterprise Oracle databases for the last five years. While you should be comfortable asking your team to adjust to a process, you should also be willing to adapt your processes to better fit your team.
3. “Waste” time. Data science projects can be like venture investing - the best projects can be 10-100x more impactful than the everyday wins. Even if they’re unlikely to succeed, data science leaders should make sure their teams have time to investigate these opportunities that have huge upsides. While it’s tempting - and often easier to justify to execs - to demand gradual progress every day, you’ll have a bigger longer term impact by uncovering just a handful of big wins.