I am a math student who has previously done a handful of regression analyses. But now I'm sitting with a classification problem and it's my first time. I wanted to know the best practices when working on a classification project.
Just like in regression it is common to fit a simple additive regression model, then with interactions, and then with non-linear expansions. After these initial steps, you can go further with more complex regression models such as ridge, lasso, random forests and so on.
Which model should I start with and which models should I try afterwards when working with classification?
FYI: I have a dataset about soil types. The response variable is soil type which has 10 different categories (sandy soil, slit soil, clay soil, ...). My task is to classify new data into these 10 different soil types from 8 other explanatory variables (deepness into the soil, hardness, ...) which are both continuous and categorical variables.
I also wanted to ask If anyone could recommend literature on classification which is math-heavy.
https://stats.stackexchange.com/search?q=machine+learning+math*+%5Breferences%5D+score%3A1$\endgroup$