Key Assumptions :
- Binary Targets : (e.g., Pass/Fail).
- Independent Observations : Observations are independent of one another.
- Low Multicollinearity : The features of the input data should not be highly correlated.
- Linearity : Linear relationship between features and the log-odds.
- Sufficient Sample Size : Maximum Likelihood demands a large sample for accurate estimation of the parameters.
- Rule of thumb : At least 10 cases with the least frequent class for each feature.