3.

Consider a classification problem where we are building a logistic regression classifier. The task is to predict if a given city has a risk of a disease epidemic or not. The data is defined using two input features or variables -

1. logarithm of size of the city
2. distance to the nearest city with epidemic

The target is a binary (0 - no risk and 1 - risk).

Consider the following data set

City #	Log Size of City	Distance	Risk?
1	2.18	0.01	1
2	1.09	18.30	1
3	-0.37	3.00	0
4	0.00	4.10	1
5	-0.42	9.00	0
6	0.09	7.20	0
7	-0.48	10.00	0
8	0.62	2.70	0
9	0.57	2.80	1
10	0.44	0.01	1

Assume that a bias term is added in the beginning. Consider the following weight vector for logistic regression: w₀ = 1.05, w₁ = -.52, w₂ = 0.85. Answer the following:

		a)	The classifier's most confident correct prediction is for city #2
		b)	The classifier's worst wrong prediction is for city #10
		c)	The classifier's worst wrong prediction is for city #9
		d)	The classifier's most confident correct prediction is for city #3