Multi class LDA was implemented from scratch as a dimensionality reduction tool and compared with Principle Component Analysis (PCA) on Iris dataset. The usefulness of LDA as a classifier was also studied and compared with logistic regression and k nearest neighbours using K-fold cross validation.
Effects of standardization was learnt. Additionally Cross validation was also used to tune the hyper parameter k of k nearest neighbours