Read the data from the csv file "spam.csv$",$ and create a dataframe object.

Build the dataset $($X$,$ y$)$ as follows: $(1)$ the first column is the class label $"$y$"($spam or ham$)$; $2$the second column is the content in each email, and transform this raw content to TF$-$IDF feature vector $"$X$".$

Randomly partition the data into training set $(60\%)$ and testing set $(40\%).$

Build the multi$-$layer perceptron $($MLP$)$ model $($design the network architecutre of MLP by yourself$)$ using training set, and apply the pretrained model to predict the class label $($spam or ham$)$ of each data point in testing set.

Analyze the predicted results using confusion matrix and evaluation metrics such as accuracy, precision, recall, f$1-$score, etc.