High-yield, short-pulse, quasi-monoenergetic neutron and charged particle experimental platforms, developed through laser direct drive implosion of thin-shell targets, are widely used in research areas such as nuclear physics, nuclear astrophysics, and high energy density physics. To achieve high neutron yield in laser direct drive implosion experiments, a data-driven high yield direct drive implosion design method is proposed. When physical understanding and numerical simulations are imprecise, this approach utilizes existing experimental data to establish empirical models and optimize parameters such as pulse shapes, capsule diameter, capsule thickness, fill pressures, laser spot diameter, etc. It effectively resolves the inherent trade-offs among implosion velocity, hot-spot symmetry, and shell-fuel mixing while balancing the relationship between one-dimensional yield and three-dimensional effects. The design process is initiated by utilizing preliminary experimental data and extensive numerical calculations to calibrate the Multi1D code. This includes refining its physical models of thermal conduction and shell-fuel mixing, as well as adjusting coefficients such as the flux limiter and diffusion multiplier. Furthermore, scaling relations of laser absorption efficiency, yield over clean (YOC), and ion temperature are derived. Subsequently, the calibrated Multi1D code, integrated with these empirical scaling relations, is employed to perform quantitative simulations to determine the optimal shell thickness and fuel pressure of the glass targets, and to predict experimental yield, ion temperature, and convergence ratio. Validation experiments using 1200 μm diameter glass targets were conducted on the 100 kJ laser facility. The results demonstrated that by applying the data-driven high yield direct drive implosion design and optimizing laser power and wavelength, a neutron yield of 1.00×10¹⁴ was achieved, representing a substantial enhancement over previous results on 100 kJ laser facility. The fusion conversion efficiency, defined as the fusion output divided by the laser energy incident on the capsule, was 4.5‰, slightly higher than the reported of NIF glass targets. It was also found that increasing laser power significantly improved implosion velocity, ion temperature and neutron yield. Additionally, the use of wavelength separation among the different cone beams effectively suppressed laser losses caused by crossed-beam energy transfer, resulting in strongly improved energy coupling efficiency, hot-spot symmetry, and neutron yield. The four-color laser configuration performed significantly better than the three-color laser. In contrast, various laser repointing strategies showed no significant impact on implosion performance, including neutron yield, ion temperature, and hot-spot symmetry.