non numerical dependent variable

#1
Hey guys!

I want to study how different variables have an effect on the type of blossom a bumblebee goes to.
Thus, my dependent variable is: type of blossom. My independent variables are: color of blossom, size of blossom, uv reflection and type of humblebee.

I haven't done anything with a non numerical dependent variable yet, so I would be really thankful for any kind of advice for how to deal with this data and which model would make sense.

Best,

Julia
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
Save them bees. I saw one when visiting Florida, but in the Midwest it had been about 30 years since I had seen a bee (excluding "sweat bees").

Multinomial regression is likely your go to. However, your question could be tricky since a bee is making a decision between what is available to them. Heck they are likely not even using their conscious brain - they are probably on autopilot based on some internal chemical signals, right. I am guessing they have no free will.
 
#3
but in the Midwest it had been about 30 years since I had seen a bee
Is this really possible? I see bees every day in the summer in Sweden, Bumble bees, many species, honey bees and solitary bees.

Multinomial distribution or number of visits as dependent variables, and colour or three variables for red green and blue, UV reflection and more.
 

hlsmith

Less is more. Stay pure. Stay poor.
#4
Well I was reference "honey bee" looking bees. I have plenty of bumblebees in little sweat bees in my back yard. Since US is not part of EU, they just nuke our whole country with insecticides, etc.

Yeah, multinomial style logistics regression or I guess a random forest could work too - depending on what she plans to do with the output.