I want to be able to input a spreadsheet where each row is a species (or whatever individual specimen), and the columns are identifying properties (flower colour, leaf size, whatever) - and then have the program produce the most efficient order/arrangement for an ID key.
So it would ideally ignore and eliminate properties that weren't useful, and would put the rest in an order that made the shortest key for the most species.
Does it exist?? Or does everyone just do it manually every time 😬
@TradescantiaHub I'm not sure I understand the problem. Can you give an example?
@peterdrake I want something that will automate those decisions for me, once I input a database which contains every identifying feature for every species covered by the key. So there would be a column for "flower colour", the value for species 1 would be "pink", the value for species 2 would be "blue", and so on - the same for every feature included. Then the program would search consider all the species and features, and decide the best order to present the questions in.
->
@peterdrake With 'best order' being based on things like, requiring the fewest identifying features to be referenced in total, and having the shortest route through the key for the most species possible.
Does that explanation help any? 🤔
@TradescantiaHub Ah, yes! It sounds like what you want is a decision tree classifier. This would likely be covered in any intro machine learning book or course.
Do you have experience with Python or R, or are you looking for a web-based tool?
@peterdrake I'm hoping for something that already exists, because realistically if it doesn't it would be quicker for me to do the job manually than automate it!
@TradescantiaHub A search on "csv to decision tree online" finds things like these:
https://datatab.net/statistics-calculator/decision-tree
Hopefully one of them is the magic you're looking for.
@peterdrake A dichotomous key is like a flow chart for identifying a species. Each step is something like "what colour are the flowers? pink -> go to 2, blue -> go to 3". Here's a random example of a dichotomous key: https://gobotany.nativeplanttrust.org/dkey/achillea/#all
Assembling one means making decisions about which identifying feature should be question 1, which should be question 2, which should be ignored/excluded from the key, and so on.
->