In Python.

$1.$ Load mnist data

$2.$ Partition data into $80\%$ training and $20\%$ testing

$3.$ Run LDA on the training set

$4.$ Project the data on a $2$d scatter plot.

$5.$ Compare scatterplot to $2$d PCA

$6.$ ensure code compiles, has proper title, x label, y label, and legend

$7.$ Answer if LDA can be projected in $3$ dimensional space. Explain why or why not.

$8.$ Make sure code actually compiles, produces plots and has no errors.

NOTE$**$ Make sure to plot both the training AND the testing set for both PCA and LDA and for $2$d and $3$d scatterplots.