3 You are given the count matrix in the table 5 for the attribute Level of rain. The class label is whether someone will decide to go out (C1) or stay home (C0). We examine the use of this attribute as our splitting attribute in an inner node of a decision tree with a binary test condition. Table 5 Level of rain ' Drizzle Light Moderate Heavy Co 1 3 6 6 C1 9 4 5 2 3.1 [2 pts] What is the classification error and the accuracy for the dataset if the classification stops at that node? 3.2 [12 pts] Which is the best way to partition the attribute values into two groups? Form the count matrices for the different partitions and compute the classification error. Evaluate each possible solution while considering the metric of classification error. 3.3 [2 pts| (i) Which is the best partition in terms of the classification error? (ii) Will you expand that node using a binary splitting criterion based on the classification error observed or will you stop expanding that node? Why