9T Questions

1. We want to predict whether a person likes a movie. We know whether the person is male or female. From a population survey we know that 60% of people like the movie and of these 20% are males. You may assume that Built. of the population is male. Use Bayes Rule to construct a simple movie prediction classifier. 2. Consider now a different movie prediction task. Suppose training data is available which consists of pairs [1m]. 1-1.2....,n. where y.-1 ii' the iIII person likes the move and ii othenvise. and iii-1 if the ill- person is male and ii othensise. Describe how to construct a logistic regression classifier for this task. 3. Logistic regression classication is often said to he most effective for linearly separable data. Explain what this means for a classifier with two inputs. What can happen is the data is no: linearly separable T 4. Download the file chdagedat from: https:ffwumscsstce}doug.leithf5'f3fcitdage.dat This contains data from a medical study. Each row corresponds to a different person and consists of three columns. The rst is the pemon ll). 11w second is their age. The third is whether they have heart disease or not [1 or ii]. |fliur aim is to use age as an inputffeature to predict whether a person has heart disease. Use logistic regression to construct a Hatlab classier for this data. and write a short report on your results [to include a plot of the original data overlaid with the classifier predictions and a report on the fraction of predictions which are correct]