Well, it's really all in how close it has to be to be called a close approach. the 2017 CA is at about 0.069 AU, or 35 times the distance to the moon.
The orbit diagram at the JPL site is only an approximation anyway, it is just a projection of the current orbit. The orbit of any object in the solar system constantly changes as it is influenced by the gravity of the Sun, (which actually orbits the center of the mass of the solar system) planets, and the larger asteroids.
For example, the 3200 Phaethon page:
http://ssd.jpl.nasa.gov/sbdb.cgi?sstr=3 ... ;cad=1#cad
Note the disclaimer on the orbit diagram:
"The applet was implemented using 2-body methods, and hence should not be used for determining accurate long-term trajectories (over several years or decades) or planetary encounter circumstances.
For accurate long-term ephemerides, please instead use our Horizons system."
In addition, the angle that you are viewing the solar system can create the illusion of close encounters when none actually occurs. If you grab the handle on the right side of the diagram, you can change the angle all the way from a direct overhead view, to one along the plane of the solar system. If you see what looks like a close approach, look at it from different perspectives.
The Horizons Ephemeris generator (link toward the bottom) on the other hand constantly recalculates the orbit using the gravitational influences of the Sun, planets, and the 3 largest objects in the main asteroid belt, which contain about half the mass of the whole belt.
The close approach table at the bottom takes all those future orbital perterbations into account, so it's far more accurate than the diagram. Again as an exmple, the orbit diagram show 0.066 AU, while the actual CA will more like 0.069 AU. Not much of a difference, but it shows that the diagram is an approximation, rather than a precise calculation.
Hope that helps.