The threshold uses the sign of the exact score. The boundary is where the score is zero, and the diagram shows that point directly.
highlighted = computed this step
The threshold rule
Class 1 is chosen exactly when z is at least 0; otherwise class 0 is chosen.
z≥0⇒1
Why the boundary is exact
The boundary is the point where z=0. Which side a point is on is determined by the sign of z.
z=0 is the boundary
Decision thresholdThe exact z sign and the z equals zero boundary determine the class decision.Classifier score tablexzdecision0-201-10201311Decision boundaryThe exact point where the score is zero.ABCDz=0
The decision on a number line
The exact scores -2, -1, 0, and 1 sit on the score axis. The cut at z=0 sends class 0 left and class 1 right, giving the same decisions 0, 0, 1, 1 the table already shows.
z≥0⇒class 1
The decision on a number lineEach exact score is placed on the score axis; the cut at z=0 splits the classes.threshold t=0 on the exact score axisclass 0 (z below 0)class 1 (z ≥ 0)-3-2-1012threshold z=0z=-2z=-1z=0z=1the score axis and the cut at z=0 are exact (σ(0)=1/2 exactly, the only exactpoint of the named sigmoid curve); the threshold is a STATED decision policyplaced by the author, NOT a learned boundary: this does not claim the cut isfitted, NOT a probability, NOT a generalization claim
Summary
The decisions 0, 0, 1, and 1 match the shown labels. No sigmoid value was needed for this exact decision.
sign(z) decides the class
Decision thresholdThe exact z sign and the z equals zero boundary determine the class decision.Classifier score tablexzdecision0-201-10201311Decision boundaryThe exact point where the score is zero.ABCDz=0