We want to use the following training set to predict whether it is suitable to play tennis?"(Play variable) based on four input variables: Outlook, temperature, humidity, windy. Instance Outlook Temperature Humidity Windy Play di Rainy Cool Normal Y NO d2 Rainy Mild High Y NO d3 Overcast Cool High Y NO d4 Sunny Hot High Y NO ds Rainy Mild High YES d6 Rainy Mild Normal N N N YES d7 Overcast Hot High YES Sunny Cool Normal IN YES ds d9 Sunny Mild High N NO Idio Sunny Hot N NO Q1: Suppose we want to build a decision tree classifier, we need to determine which attribute should be used first to split the root node a) Briefly specify how can we use Gini Index to find the best split. b) If we use Windy attribute to split the data set into two subsets (If Windy=Y, place the data point in node 1; If Windy=N, place the data point in node 2), calculate the Average weighted Gini Index after the split. Please try to be as detailed as possible. Q2: Suppose we want to classify the following new instance d11 using KNN method. Instance Outlook Temperature Humidity Windy d11 Sunny Cool Normal IN a) What is the distance between instance di and instance d11? If we set k=1, what outcome do we get in classifying this new instance? Why? Please try to be as detailed as possible. Hint: use nominal distance to calculate the distance between each instance in the training data and instance d11. Find the nearest neighbor of d11. We want to use the following training set to predict whether it is suitable to play tennis?"(Play variable) based on four input variables: Outlook, temperature, humidity, windy. Instance Outlook Temperature Humidity Windy Play di Rainy Cool Normal Y NO d2 Rainy Mild High Y NO d3 Overcast Cool High Y NO d4 Sunny Hot High Y NO ds Rainy Mild High YES d6 Rainy Mild Normal N N N YES d7 Overcast Hot High YES Sunny Cool Normal IN YES ds d9 Sunny Mild High N NO Idio Sunny Hot N NO Q1: Suppose we want to build a decision tree classifier, we need to determine which attribute should be used first to split the root node a) Briefly specify how can we use Gini Index to find the best split. b) If we use Windy attribute to split the data set into two subsets (If Windy=Y, place the data point in node 1; If Windy=N, place the data point in node 2), calculate the Average weighted Gini Index after the split. Please try to be as detailed as possible. Q2: Suppose we want to classify the following new instance d11 using KNN method. Instance Outlook Temperature Humidity Windy d11 Sunny Cool Normal IN a) What is the distance between instance di and instance d11? If we set k=1, what outcome do we get in classifying this new instance? Why? Please try to be as detailed as possible. Hint: use nominal distance to calculate the distance between each instance in the training data and instance d11. Find the nearest neighbor of d11
