Create arbitrary decision tree without training data?

3 ビュー (過去 30 日間)
Alex van Tuyll
Alex van Tuyll 2019 年 10 月 18 日
コメント済み: Alex van Tuyll 2019 年 10 月 31 日
Hi there,
I am modelling rule-based control in greenhouses (with lots of rules like "if temperature > 25°C, open windows"). This essentially means making a decision tree.
Having looked around, I like the functionality of MATLAB's fitctree() classification function. I like that the trees can be visualised and used elegantly. However, the type of tree I want to create does not revolve around data. I want to build a tree with an arbitrary structure and arbitrary threshold values myself.
Does MATLAB have anything that can do this? Input arguments would be variable names (e.g. temperature) and threshold values (e.g. 25) - in a certain order - and the output would give a tree with the same functionality as fitctree() trees.
Thanks and best wishes,
Alex
PS - I see phytree() does something similar, but can it be used for decisions with multiple variables, and can it give a decision using the current state variables as an input?

回答 (1 件)

Asvin Kumar
Asvin Kumar 2019 年 10 月 29 日
  1 件のコメント
Alex van Tuyll
Alex van Tuyll 2019 年 10 月 31 日
Thank you, Asvin! Looks like that will help with the visualisation aspect especially.

サインインしてコメントする。

カテゴリ

Help Center および File ExchangeElectromechanical についてさらに検索

製品


リリース

R2018b

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by