I'm currently trying to analyse a dataset containing multiple non-ordinal categorical features and a binary target variable. The table looks something like this:
+------------+---------+------------+--------+
| Col1 | .... | Col14 | Target |
+------------+---------+------------+--------+
| cat 1 | cat 1 | cat 1 | 0 |
| ... | ... | ... | ... |
| cat 9 | cat 50 | cat 450 | 1 |
+------------+---------+------------+--------+
The entire table is 400.000 rows x 15 columns, from which the last column is the target variable. Each feature has multiple non-ordinal categories ranging from 9 categories to multiple hundreds of categories.
My first instinct would be to one hot encode all the categorical variables. However, I'm scared that doing so will make any model prone to overfitting.
How could I handle/encode the features variables to analyse their effect on the target variable, using Python?