You have a discrete graph if data on one of the axes can take only discrete values, e.g., the count of how many times each side of a dice turns up when you roll it a few dozen times. Your horiz axis needs discrete numbers 1-6 and it doesn't make sense to want to interpolate to find how many times the graph predicts that the number 2.8 would come up. So you don't join the plotted points because a continuous plot tacitly invites readers to interpolate between the plotted points.

 📅 📆 🎡 🎲 🎰