Coupling the magnetosphere with the interplanetary medium, magnetic reconnection at the magnetopause serves as the main driver of magnetospheric activity. The efficiency of this coupling is highly contingent upon the location at which reconnection occurs on the magnetopause. Although various models have been proposed to predict the location of reconnection lines, also referred to as X-lines, none of these models are consistently aligned with global MHD simulations. These models often rely on quantities whose spatial distributions at the magnetopause are obtained by analytical or numerical modeling. To study and constrain magnetic reconnection on a large scale from an observational standpoint, we used statistical learning to automatically select in-situ measurements from four missions (Cluster, Doublestar, THEMIS, MMS). This allowed for a 3D reconstruction of the magnetic field draping in the dayside magnetosheath, and the discrepancies with a commonly used magnetostatic model revealed the significance of the plasma flow in the draping mechanism. It also enabled the investigation of the impact of magnetic reconnection on the magnetic pileup in the magnetosheath, on which previous studies have not been consistent. Finally, the reconstruction of the global distribution of magnetic shear angle, current density, and asymmetric reconnection rate at the dayside magnetopause provide some constraints on the location of magnetic reconnection. The location of the X-line on the dayside magnetopause will be further discussed in relation to both global and local modeling.