Abstract: Spatial distribution modelling of invasive species is crucial for deepening the understanding, predicting, and managing of biological invasions, in order to provide a scientific foundation to effectively address this challenge.In doing so, providing reliable pseudo-absence species (negative samples) is one of the core components of spatial distribution modelling of invasive species.Based on 124 invasive samples of Canada goldenrod (Solidago canadensis L.) and 11 climate variable datasets within the Yangtze River Economic Belt, this study employed cosine similarity to calculate the association between candidate negative samples and invasive species.The Getis-Ord Gi* statistical method was integrated to generate z-score variables for measuring spatial autocorrelation.A framework for identifying pseudo-absence samples of invasive species, by considering climate similarity and spatial correlation, was constructed to reveal potential suitable areas for invasive species.The results indicated that: (1) Compared to previous studies, the pseudo-absence samples generation method presented in this study demonstrated a superior performance in model prediction, feasibility and effectiveness.(2) The pseudo-absence sample sampling strategy helped address the challenge of potential invasive points that might be mis-sampled due to random sampling.This sampling strategy was able to identify suitable areas for invasive species of different levels.(3) Apart from the northwest part of Sichuan Province, most areas within the Yangtze River Economic Belt were suitable for the growth of Canada goldenrod, especially in Shanghai, Jiangsu, Zhejiang, and Anhui.Therefore, it was necessary to pay focused attention, and to actively share experiences of joint prevention and control measures.