When to Reach for a Tree
Last timeWhich Input Actually Mattered
On data that lives in a spreadsheet a committee of trees usually wins, on data made of pixels, sound or words it usually loses, and the shape of the data says which case you are in.
The claim, stated plainly
On data that arrives as rows and columns with names on the columns, a
committee of trees is usually the most accurate thing available and takes a
few minutes to fit. On data that arrives as pixels, audio samples or text, a
network wins and nothing in this course comes close.
That division is sharp enough to be useful as a rule, and the reason for it is
specific rather than mysterious.
Why the division falls where it does
A split asks about one column against a threshold, which presumes that the
column means something on its own. Income above thirty thousand is a
meaningful statement. That is the usual situation in a spreadsheet, where
somebody chose and named each column because it meant something.
Now consider a photograph. The value at pixel four thousand and twelve means
nothing on its own; it means something only in relation to its neighbours, and
the same object appearing twenty pixels to the left is an entirely different
set of column values. A tree would have to rediscover that relation one
question at a time, and it cannot.
| a single input means som | which family wins | typical gap | |
|---|---|---|---|
| spreadsheet columns, mix | 1 | 1 | 3 |
| sensor readings over tim | 1 | 1 | 1 |
| pixels of an image | 0 | 2 | 40 |
| audio samples | 0 | 2 | 50 |
| words of a document | 0 | 2 | 35 |
The gaps are asymmetric and that matters for the decision. Choosing a network
for spreadsheet data costs a few points and a great deal of time. Choosing a
tree for images costs everything.
The lesson stops here
6 more paragraphs to go
You have read the opening. The rest of the argument, the problems that check whether it landed, and the lines worth keeping at the end all come with a plan.
The first lesson of every course in the library reads the whole way through, free, so you can see exactly what the rest of them are.
See the planThe contentsThis is the reading half
Starting the course gives you your own copy of it. Every idea on every page has problems standing under it, marked with a reason rather than a tick, and any sentence you do not believe can be opened and argued with. None of that can happen on a page nobody owns.
The contents