Applied Data Science Partners interview question

Code review: - Use of one-hot encoding over ordinal encoding - fixing the accuracy, precision, recall calculation, and explaining them - dropping the target before training the model - fitting the model on the train dataset instead of test - explaining the choice of model – random forest classifier, and the number of estimators i.e. trees (hyperparameter) - Why use validation dataset and then predict the test dataset? - Label mismatch in the train and test dataset (additional column in the train dataset) - Missing values in ‘age’ feature - How to handle missing values?