Help out your future self

Help out your future self

I’ll be honest with you… I actually really enjoy building flat pack furniture. The step-by-step approach appeals to my way of working (I count out all the pieces before I start) and there’s a definite sense of achievement when you get the final product. However there is a situation where it’s less fun and becomes more stressful. This happens when there is ambiguity in the instructions - I’m sure you’ve been there. Exactly which screw is it? Which way up does this piece go? Have I missed a step??

There is a really clear link to how you think about the documentation of your model building process. Having a well written document of your steps will help you prove to regulators that you know what you're doing, make it easy to replicate the process if needed and answer all those quick questions that pop up over the lifetime of the model. Not only will your future team thank you for this, so will your future self!

So what makes good model documentation? Well firstly your documentation should not be Lord of the Rings. You should avoid telling the long twisty story of the last 6 months and instead focus on the key facts of the build: why you built the model, how you defined your data, how you built the model etc. Secondly I find it really useful to have a clearly agreed set of standards for where your code, data and charting can be found making it easy to manage and reference. Finally, where possible write your documentation in a way that makes your model easy to reproduce. If you do this you won’t spend the next 5 years wondering: where did I get that data from or how did I transform this feature? One big difference from the flat-pack instructions is that great model documentation also talks about the ‘why’ as well as the ‘what’. Make sure you not only explain the decisions but the rationale behind those decisions.

In summary, try to think about your documentation from the perspective of you in 5 years time. What would you have wished you’d written down? Also make sure you have committed an appropriate amount of time to complete this - it shouldn’t be an afterthought and is as important as the actual model equation.

Questions to ask:

  • What are the key elements of your model build project that you need to write down?
  • How will you manage version control and reference datasets and code?
  • What is your approval process to ensure documentation is complete and makes sense?

Last week: what can go wrong... and what will you do about it?

Next week: opening up the watch

Sam Birkinshaw

Passionate Senior Programme Manager - Change Delivery Leader | Women in Tech Coach + Mentor | OKR Coach | Aerospace, Financial Services, Technology

2 年

I quite enjoy building flat pack too - good analogy!

回复
Satish Patil

Building Kanlet

2 年

Nice read

回复

要查看或添加评论,请登录

Dan Kellett的更多文章

  • My 4 microblogs on AI governance

    My 4 microblogs on AI governance

    Over the last 4 weeks I have looked to cover key learnings from my 21 years being involved in the governance of Machine…

    1 条评论
  • Data karma

    Data karma

    AI success relies on a large amount of knowledge. This may be technical knowledge, data knowledge or business knowledge.

    2 条评论
  • Goldilocks and SQL

    Goldilocks and SQL

    Last week I wrote about my early years as a data scientist and the challenge of jumping the experience chasm as I moved…

    2 条评论
  • Wise council

    Wise council

    I joined Capital One straight out of university. I completed my Bachelors degree in Mathematics and Statistics and…

    1 条评论
  • The Jets and the Sharks

    The Jets and the Sharks

    This week I want to tell you a story about one of my earliest model building projects. I was a recent graduate making…

    1 条评论
  • My 8 microblogs on AI model building

    My 8 microblogs on AI model building

    Over the last 8 weeks I have looked to cover key learnings from my 21 years building Machine Learning models in…

  • Occam’s Razor

    Occam’s Razor

    Buying a new car can be a pretty daunting experience unless you know exactly what you want. Deciding on a make and…

  • Opening up the watch

    Opening up the watch

    Imagine it’s your birthday and there’s a knock on your door. The delivery person hands you a beautifully wrapped parcel…

    1 条评论
  • What can go wrong... and what will you do about it?

    What can go wrong... and what will you do about it?

    Sometimes, despite everyone’s best intentions, things go wrong. Good risk management can help ensure this doesn’t have…

    3 条评论
  • But is it working?

    But is it working?

    Sometimes it can feel like a long slog building and deploying a new AI application. The process of defining the…

    1 条评论

社区洞察

其他会员也浏览了