There’s this wonderful article I want to share on building prediction models using ensembles. “Ensembles” in this case simply means the combination of two or more prediction models.
I’d personally had great success bringing several (relatively) poorly performing models together into one ensemble model, with prediction accuracy far greater than any of the models individually.
Definitely something to check out if you’re into this sort of thing.
The following passage is taken from the beautiful book Master of the Three Ways by Hing Ying Ming (which libraries might classify as “Eastern Philosophy”):
The net is set for the fish,
But catches the swan in its mesh.
The praying mantis covets its prey,
While the sparrow approaches from the rear.
Within one contrivance hides another;
Beyond one change, another is born.
Are wisdom and skill enough to put your hopes on?
Just a little reminder for my future self on the uncertainty of life (r-squared never is 100%).
Update: For the uninitiated, my comment on “r-squared” above was just a little statistical quip. R-squared is a number between 0 and 1 that represents the amount of variability of a linear model, in percentage, that can be explained by the model. Anything outside of r-squared, so 1 less r-squared, is uncertainty.
Seth Godin wrote a wonderful post on how we sometimes need an external push (through laws, policies, cultural guardrails) to do what’s best for us. It can be basically summed up by the following statements (from the post):
- We know that wearing a bicycle helmet can save us from years in the hospital, but some people feel awkward being the only one in a group to do so. A helmet law, then, takes away that problem and we come out ahead.
- Guard rails always seem like an unwanted intrusion on personal freedom. Until we get used to them. Then we wonder how we lived without them.
I was just thinking about true this is for so many other aspects of our lives. The friends we choose, because of the context they set, determine many of the decisions we make, and consequently many of the paths of life we take.
When setting up a company, a department, a team – how important it would be then to make sure that the cultural norms we encourage and enforce are the ones we want.
Whether it’s a culture of success (however you define it); freedom of experimentation; openness of communication; risk taking; or hard work, it is our job as servant leaders to ensure that it’s the least awkward thing to do.
There is a huge difference in working with a team that you know will be with you for only a single project and a team that you know will be with you for still many more.
When you’re working with a team that you know will be with you for a long time, you may do what needs to be done to achieve a favourable outcome for this project, but understand the outcome of the sum total of all potential projects to come is just as important, if not more: setting the right precedent and ensuring goodwill among all (as far as is reasonable!) needs just as much attention.
But with the team that you work with for only a single project, you do what needs to be done to achieve the best outcome for this project without too much regard to how that might implicate future interactions with the team. Thinking long-term when you shouldn’t could potentially hurt the outcome of this project.
If you’re only going to be working with them for this one time, setting a bad precedent or upsetting one or another doesn’t matter too much.
(This post is more a reminder to me than anything else: last year I worked on several one-off projects, during which I was always in this “long-term” mode of thinking. I tried pleasing everybody and making sure I didn’t set poor precedents – “fairly” distributing workload, for example, to people whom I knew couldn’t perform, ultimately hurting the results of these projects.)
I’ve written about shipping before: the act of delivering a product; an article; a report; a piece of art. You can have the best ideas in the world, but if you don’t ship, they’re worth as much a ton of gold at the bottom of a rubbish heap.
“We don’t know if the data’s 100% right – are you sure we should publish it? What if they question us? What if we have to change something later? Shouldn’t we validate some more till we’re completely sure?”
Yes, you should – if you had all the time in the world.
But we don’t.
We have done our homework; we know the assumptions; we know there are issues with the data but these are not show-stoppers. For our purposes, 90% is good enough. If we waited till we were 100% before shipping, nothing would be shipped.