This research presents a comparative study of specialized deep learning and state of the art machine learning approaches for multiclass classification of six emotions: joy, sadness, surprise, fear, love, anger. Specialized deep learning algorithm Bi-LSTM; state of the art H2O ANN; traditional SVM and state of the art H2O Gradient Boosting Machines are applied on vectorised text. Cross validated performance using different vectorization techniques: text to sequences, word to vectors and TF-IDF are presented. The specialized Bi-LSTM on text to sequence vectorised data outperforms SVM and both outperform H2O ANN. A Gradient Boosted Machine age classifier is used to stratify test data. The traditional TF-IDF applied SVM outperforms the Bi-STM model on both Youth and Adult test data. The research is further extended to present a chatbot deployment of the emotion classifier.