## First, Create a Multilayer Perceptron Classifier $($Neural Network$)$ to classify loans as accepted or rejected $($Y or N$).(15$ pts$)$

$-$ Drop Gender, Loan$\_$ID$,$ and Missing Values before training.

$-$ We will inspect Gender after training, so be careful how you drop it before training and testing.

$-$ Hint: Wait until scaling and create new variable name, $`$X$\_$train$\_$sc$`,`$X$\_$test$\_$sc$`,$ to save $`$X$\_$train$`,`$X$\_$test$`$ for later.

$-$ Train$/$Test with a $80/20$ split and stratify on Gender.

$-$ Remember to drop outliers from $`$ApplicantIncome$`,`$CoapplicantIncome$`,$ and $`$LoanAmount$`.$

$-$ Remember to create integer dummies for categorical variables.

$-$ Even if a variable is numeric, it may still just be a category. Check with $`.$value$\_$counts$()`$

$-$ Remember to scale the data without violating data leakage

$-$ Create your Classifier with at least $3$ Layers. You are free to choose how many nodes are in each layer.

$-$ To the best of your ability, search for a stable model. You will not be graded on model performance.

$-$ Display the Confusion Matricies for both the Train and Test sets.

$-$ Display the Feature Importance for the Training set. Is this expected?

$-$ Comment on the Stability of the Model.