Which is not a rule for read$\_$csv$()$ as related to numbers:

If any value of a column is non$-$numeric, the column's values are represented as Python types as opposed to NumPy types

If all values in a column are numeric and any value has decimal places, the column is represented as a NumPy float

if any value in a column of numbers is missing, even is all non$-$missing values are integers, the column is stored as a NumPy float type

If a column has integer values and some missing values, pandas will use the Python int type or None for the values in the column