The installation of photovoltaics (PV) solar systems globally is projected to become the highest renewable energy source by 2029, with Sweden alone targeting 8 GW of installed capacity by 2030, creating an urgent need for intelligent and scalable spatial planning tools. However, Sweden lags in utility-scale solar energy development, as the distributed solar PV accounts for more than 90% of total installed PV capacity. Predominant site selection methods rely on multi-criteria decision analysis applied within geographic information systems (GIS), or combined with power flow models. Thus, it lacks historical approval utility scale patterns and data on existing installed parks, which limits their ability to predict future solar park accurately. This study proposes an integrated framework combining GIS with a machine learning algorithm, trained on real-world data from existing and approved solar parks, where suitable locations are first identified based on physical and regulatory criteria, ensuring that only spatially realistic sites are selected. Input features include solar irradiance, suitable, unsuitable, protected nature, national-interest, and forest-specific unsuitable layers, together with restricted and buffered layers accounting for infrastructure, terrain, irradiation, and proximity constraints, where the model output is a binary suitability classification for each location. The model can help county administrative boards to identify where future solar parks have the best potential to be established. The model would also guide developers in their decision-making and visualize potential locations on a map, to support Sweden in achieving its solar energy target by 2030. Future work will extend the framework by introducing a financial viability assessment layer dependent on land use and electricity prices, evaluating identified sites as standalone PV parks and as hybrid PV-battery storage systems to further enhance project viability under realistic market conditions and resources assessment.