Abstract: Winter wheat is one of the most important grain crops in China, and its yield is of great significance to national food security. The yield of winter wheat is jointly influenced by multiple factors, including genotypic characteristics, growing environments, soil conditions, growth performance across all developmental stages, and field management practices; suboptimal growth at any stage can compromise the final yield. This study aimed to develop a comprehensive scoring model for winter wheat varieties based on multi-source data. Correlation and multicollinearity analyses were conducted on phenotypic factors and environmental adaptability factors across growth stages to remove redundant factors and retain core information. Principal component analysis (PCA) was then employed to extract principal components, and to generate scores and rankings for the tested varieties. Thirty-one key cultivated winter wheat varieties from Lingcheng District, Dezhou City, Shandong Province were selected as experimental materials. Eleven phenotypic factors and 12 growth rate factors were measured and collected throughout the entire growth period. The regularized Mahalanobis distance-improved Pearson model was adopted to analyze inter-factor correlations and quantify the mutual influence between factors. Potential multicollinearity risks among factors were identified based on the correlation coefficients. The improved Pearson model significantly enhanced its robustness against outliers in agricultural small-sample datasets. On this basis, multicollinearity analysis was performed using tolerance (TOL), variance inflation factor (VIF) and independent information content. Redundant factors with overlapping information or high correlation were excluded, and key input factors for PCA were obtained. Finally, principal components were extracted via PCA, and a principal component scoring formula was established to calculate variety scores and generate rankings. The improved Pearson model effectively mitigated the interference of outliers. When the outlier proportion reached 15%, the mean absolute error (MAE) of the model was 0.07, and the mean square error (MSE) was 0.008. Correlation analysis revealed that the absolute values of correlation coefficients among some factors exceeded 0.8, indicating potential multicollinearity risks. By integrating TOL, VIF and independent information content, 8 key factors were ultimately screened as input variables for PCA. The PCA model successfully extracted 4 principal components (with a cumulative variance contribution rate exceeding 99%) and comprehensively scored the 31 winter wheat cultivars. The top 5 cultivars ranked by comprehensive score were Luyan 373, Luyan 951, Jimai 33, Jimai 34, and Jimai 44. The comprehensive scoring model systematically quantified the yield potential and stability of each cultivar from the perspectives of static growth performance and dynamic growth rates across different developmental stages of winter wheat. Its scoring results can serve as a reference for short-term regional winter wheat breeding and screening. In the future, after multi-point and multi-year experimental verification, it can provide a more reliable decision-making tool for long-term intelligent wheat breeding and extension in the region.