Computer Science and Engineering era have resulted in robust and powerful artificial intelligence special computer products-chatbot having capability of intelligent auto responses. Optimisation chatbot relies on optimal model, with challenging task of selecting optimal hyper parameter values. Optimised chatbot performance depends on hyper parameter tuning which resolves the challenging task by selecting the best hyper parameters. In this work, we design and develop 7 different machine learning chatbot models decision tree, random forest, logistic regression, extra tree, AdaBoost, Gradient Boost, support vector machine and optimise chatbot with technique of grid search hyper parameter optimisation. Tenfold cross validation is used to avoid bias. The performance of these 7 optimised chatbot are compared with factor: fitting time, accuracy, precision, recall, f1-score, Gini coefficient, MAE, MSE, RMSE, AUC ROC, R-Squared and execution time for considered hyperparameter search space, to select best machine learning classifier for optimised chatbot. Best values of hyperparameters obtained in our experiment for seven different models have high impact on enhancing performance of chatbot. Behind every successful chatbot optimisation, time complexity of algorithm plays a vital role. Results predict best classifiers for optimised chatbot based on specific performance factors. If optimised chatbot needs low fitting time and low execution time then extra tree is best classifier as per the requirement. Best classifier with high accuracy and f1-score is logistic regression. For 6 best performance factors high precision, ROC AUC, Gini coefficient and low MAE, MSE, RMSE then optimised chatbot best classifier is Gradient Boost. Other best classifiers for optimised chatbot are decision tree with high recall, low RMSE, AdaBoost with low R-squared, random forest with high accuracy and support vector machine with high accuracy.