The magnetic component is responsible for transmitting,storing,and filtering magnetic energy,which directly affects the volume,weight,loss and cost of the power converter. Therefore,accurately predicting magnetic core loss is particularly important. To address the issue of inaccurate core loss evaluation in magnetic components,a data-driven core loss modeling method is proposed. First,decision tree and eXtreme Gradient Boosting (XGBoost) models are used to classify excitation waveforms and the flux density distribution and waveform characteristics of each material in the test set are plotted. Second,models based on XGBoost,support vector machine,gradient boosting regression tree and K-nearest neighbor are established to predict the core loss of samples in the test set. Then,single objective optimization models based on genetic algorithm and particle swarm optimization,as well as a multi-objective optimization model based on Non-dominated Sorting Genetic Algorithm Ⅱ(NSGA-Ⅱ) algorithm,are proposed to obtain optimal conditions such as temperature,frequency and waveform parameters corresponding to the best objective function values. The results show that XGBoost performs best in both waveform classification and core loss prediction,with prediction accuracies of 85.66% on training set and 84.83% on test set,respectively. The NSGA-Ⅱ algorithm achieves the best performance in the joint optimization of core loss and transmitted magnetic energy.