3 habits of highly effective analysts

Most analysts start out with a good deal of enthusiasm for coding and clever sounding mathematical models. I certainly did. But to become a good and valuable analyst, you need to hone some of the softer skills.

I know one or two analysts who seemed to start out with these habits. They were far more interested in understanding the client and solving their problems than in becoming a wiz kid at Python or R. I was not one of them!

But gradually over years of working on lots of varied client projects and working with lots of talented people, I think I’ve picked up these habits.

I don’t think they can be learned by reading a book (or a blog post)…But at least if you know what you’re looking out for, you might pick them up faster.

So, here is my list:

1. Empathy for the stakeholder

  • Solving their problems, not the problems you want to solve

  • Find out what their key decision points and constraints are

  • You don’t need to make your models simple, but you do need to be able to explain what they’re saying in simple terms

2. Check, check and check again

  • Check the data. Are there duplicates? Are there strange patterns over time. Are you absolutely sure you’re aggregating in the right way? Are your dates right? Are you missing anything? Are you introducing errors when joining datasets?

  • Check your assumptions. Write them down. Talk it through with someone or if you can’t, write down in long form what you’re doing and why.

  • Check your results. Get a feel for what the numbers should be before running any models. Sensitivity test - what happens if you change your model a little or exclude some data points.

3. Be tidy

  • Even if you aren’t going to be sharing your code with someone

  • IMO this is harder in Python / notebook environment than R (if you’re using tidyverse!). It takes extra effort to make notebooks tidy, but it’s worth doing

  • Directory structures, white spaces in code, version control, naming conventions you stick to

Got any others? Let me know!


About me

I am a data / marketing scientist and statistician, with 15+ years experience building models to solve problems. My focus areas are Bayesian modelling, customer analytics, causal inference and simulation. Client list includes Scribd, the Economist, ITV and Elvie.

I’m always keen to discuss and debate things statistical, especially Bayesian and especially MMM - so get in touch!

Previous
Previous

How to measure customer lifetime value

Next
Next

Is Bayesian MMM worth the faff?