This work demonstrates a novel approach to design and optimization of rare-earth free magnetic materials for targeted properties by effectively using various computational and statistical tools. From the open literature, we defined the alloying elements and bounds of their concentrations to develop a new system of Alnico alloys. Initial compositions of candidate alloys were generated using a quasi-random sequence generation algorithm. Response surface methodology approach was used to develop surrogate models to efficiently link alloy chemistry with desired macroscopic properties for these multi-component systems. The most accurate meta-models were used for multi-objective optimization of desired properties by utilizing various evolutionary approaches. Various statistical tools and pattern recognition techniques were used to determine patterns and correlations within the created dataset. Pareto-optimized candidate alloys were experimentally validated and used to improve the accuracy of the response surface generation used by the multi-objective optimizer to find the next generation of Pareto-optimal alloys. Results over the cycles show significant experimentally verified improvement in the properties of these alloys.