We measure the dependence of planet frequency on host star mass, M L, and distance from the Galactic center, R L, using a sample of planets discovered by gravitational microlensing. We compare the two-dimensional distribution of the lens-source proper motion, μ rel, and the Einstein radius crossing time, t E, measured for 22 planetary events from Suzuki et al. with the distribution expected from Galactic model. Assuming that the planet-hosting probability of a star is proportional to MLmRLr , we calculate the likelihood distribution of (m,r). We estimate that r=0.10−0.37+0.51 and m=0.50−0.70+0.90 under the assumption that the planet-hosting probability is independent of the mass ratio. We also divide the planet sample into subsamples based on their mass ratio, q, and estimate that m=−0.08−0.65+0.95 for q < 10−3 and 1.25−1.14+1.07 for q > 10−3. Although uncertainties are still large, this result implies a possibility that, in orbits beyond the snowline, massive planets are more likely to exist around more massive stars whereas low-mass planets exist regardless of their host star mass.