Design the steps involved in PCA to reduce the dimensionality of MNIST dataset wth n$\_$componenets as $2.$ Then train the logistic regression model on reduced dataset and print its accuracy and confusion matrix.