Abstract: To rapidly determine anthocyanin content in red radish (Raphanus sativus L.) and improve raw material screening efficiency for pigment production, color parameters (L^*, a^*, b^*, C^*, and H^*) were measured at different positions on red radish using a colorimeter. Ultra-performance liquid chromatography-mass spectrometry was employed to analyze content of individual anthocyanin. A significant positive correlation was observed between a^* values and anthocyanin contents. Linear regression models based on a^* values achieved a 68% prediction accuracy for total anthocyanin content and from 28.9% to 52.6% for major individual anthocyanins. Eight machine learning algorithms were employed to optimize the color parameter-based prediction model of anthocyanin content. The Lasso regression model demonstrated strong predictive performance for total anthocyanin content, achieving a prediction accuracy of 87.5%. The Decision Tree regression model achieved prediction accuracies of 75% for both pelargonidin-3-(feruloyl)diglucoside-5-(malonyl)glucoside and pelargonidin-3-(caffeoyl)diglucoside-5-glucoside, indicating its effectiveness in estimating specific anthocyanin compounds. Collectively, these findings indicate that combining colorimetric measurement with machine learning enables rapid prediction of anthocyanin content, which is conducive to raw material selection in red radish pigment production.