Advancing Machine Learning Interpretability: LIME Applications to Temporal Data and A Novel Method of Counterfactual Generation