Assuming that we are using KNN classification algorithm to train a model that predicts whether a game will likely be generating high vs low rewarded video ad revenue.

Input features are:

$(1)$ binary coded data indicating presence or absence of game features $($e$.$g$.,$ daily mission, collection system, quests, tournament, guild system$),$

$(2)$ game$$s daily in$-$app purchase revenue per daily active user,

$(3)$ game$$s daily active user,

$(4)$ game$$s percent of daily active user in different regions,

$(5)$ game genre,

$(6)$ users$$ daily active time average $,$

$(7)$ day $30$ retention,

$(8)$ number of different game modes that the game has

$1.$Describe any possible data preprocessing steps for the input data

$2.$Describe how you would choose the hyperparameters for the KNN model

$3.$ Describe the training process

$4.$Describe which loss function will be appropriate and why

$5.$ Describe the method for selecting the best model $($which quality metrics will be used$)$

$6.$Describe how will iterative model and data update be done

$7.$ Describe how you can track model prediction quality when it is rolled out to using real data and how can the model be improved in the future