We develop a method based on the angular distribution on the sky of cosmological parameters to probe the inhomogeneity of large-scale structure and cosmic acceleration. We demonstrate this method on the largest type Ia supernova (SN) data set available to date, as compiled by the Joint Light-curve Analysis (JLA) collaboration and, hence, consider the cosmological parameters that affect the luminosity distance. We divide the supernova sample into equal surface area pixels and estimate the cosmological parameters that minimize the chi-square of the fit to the distance modulus in each pixel, hence producing maps of the cosmological parameters $\{\Omega_{M}, \Omega_{\Lambda}, H_{0}\}.$ In poorly sampled pixels, the measured fluctuations are mostly due to an inhomogeneous coverage of the sky by the SN surveys; in contrast, in well-sampled pixels, the measurements are robust enough to suggest a real fluctuation. We also measure the anisotropy of the parameters by computing the power spectrum of the corresponding maps of the parameters up to $\ell=3.$ For an analytical toy model of an inhomogeneous ensemble of homogeneous pixels, we derive the backreaction term in the deceleration parameter due to the fluctuations of $H_{0}$ across the sky and measure it to be of order $10^{-3}$ times the corresponding average over the pixels in the absence of backreaction. We conclude that, for the toy model considered, backreaction is not a viable dynamical mechanism to emulate cosmic acceleration.