5) (35 marks) Submit your solution as a file named a2q5.txt or a2q5.pdf using the command submit-nlp. Let us assume that you work on a problem of detecting positive microblog messages about financial markets. After analyzing a set of messages you found that three features Fi, F2, and F3 of messages are particularly useful in recognizing whether a message is positive or negative. You decided to work on creating a small Nave Bayes classifier to classify messages into positive and not-positive classes. To summarize, your model uses the following features: The feature Fi E {t, f}, which is set to 't' (true) if feature Fi is present in a message, and otherwise it is set to 'f' (false). The feature F2 E {t, f}, which is set to 't' (true) if feature F2 is present in a message, and otherwise it is set to 'f' (false). The feature F3 {t, f}, which is set to 't' (true) if feature F3 is present in a message, and otherwise it is set to 'f' (false). The class itself is represented using the class variable CE {p, n}, where p stands for a positive message, and n stands for a non-positive message. The training data is presented in the following table: messages - * - a) (15 marks) Calculate the conditional probability tables (CPI for the Nave Bayes model. b) (5 marks) Calculate P(C = p Fi = f, F2 = t, F3 = f) using the Nave Bayes model and briefly describe what this conditional probability represents. c) (5 marks) What is the most likely value of the class variable C for the partial configuration (Fi = f, F2 = t, F3 = f) according to the Nave Bayes model discussed in a) and b)? d) (5 marks) What is P(C = p F1 = f, F2 = t, F3 = f) if we use the Joint Distribution Model? - * = = = = = = = 9444 - - * * e) (5 marks) What is P(C = p Fi = f, F2 = t, F3 = f) if we use the Fully Independent Model? f ftp fft In 90 21 Note: In assignments, always include intermediate results and sufficient details about the way the results are obtained. 5) (35 marks) Submit your solution as a file named a2q5.txt or a2q5.pdf using the command submit-nlp. Let us assume that you work on a problem of detecting positive microblog messages about financial markets. After analyzing a set of messages you found that three features Fi, F2, and F3 of messages are particularly useful in recognizing whether a message is positive or negative. You decided to work on creating a small Nave Bayes classifier to classify messages into positive and not-positive classes. To summarize, your model uses the following features: The feature Fi E {t, f}, which is set to 't' (true) if feature Fi is present in a message, and otherwise it is set to 'f' (false). The feature F2 E {t, f}, which is set to 't' (true) if feature F2 is present in a message, and otherwise it is set to 'f' (false). The feature F3 {t, f}, which is set to 't' (true) if feature F3 is present in a message, and otherwise it is set to 'f' (false). The class itself is represented using the class variable CE {p, n}, where p stands for a positive message, and n stands for a non-positive message. The training data is presented in the following table: messages - * - a) (15 marks) Calculate the conditional probability tables (CPI for the Nave Bayes model. b) (5 marks) Calculate P(C = p Fi = f, F2 = t, F3 = f) using the Nave Bayes model and briefly describe what this conditional probability represents. c) (5 marks) What is the most likely value of the class variable C for the partial configuration (Fi = f, F2 = t, F3 = f) according to the Nave Bayes model discussed in a) and b)? d) (5 marks) What is P(C = p F1 = f, F2 = t, F3 = f) if we use the Joint Distribution Model? - * = = = = = = = 9444 - - * * e) (5 marks) What is P(C = p Fi = f, F2 = t, F3 = f) if we use the Fully Independent Model? f ftp fft In 90 21 Note: In assignments, always include intermediate results and sufficient details about the way the results are obtained