Recent years have witnessed the booming development of carbon market concerning the global environment changes. To address the concern, the carbon credit price is considered as an effective tool in practice. However, the carbon price changes in most markets exhibit nonlinear, non-stationary, and highly volatile characteristics. It poses significant challenges to construct effective data-driven models to learn from these data. Currently, most existing models consider historical carbon trading data as the singular input, which cannot effectively capture price trends during significant fluctuations. Moreover, existing models do not fully leverage the various variables as the input. Therefore, in this work, we present a novel hybrid model, termed as MIF-CP, which takes multiple influencing factors of carbon prices as input. The model incorporates historical carbon trading data and external influencing factors, such as economic factors, energy prices, and price fluctuations in other carbon markets. We introduce modified ensemble empirical mode decomposition method to extract the latent relationship from these data, following with correlation coefficient method for feature selection across different variables. The model is based on Kernel Extreme Learning Machine (KELM), and we incorporate the Whale Optimization Algorithm (WOA) to optimize the target function. Experimental results show that our hybrid model achieves the best performance in comparison with both single models and hybrid models from the literature in terms of prediction accuracy, and it is also reliable in terms of runtime. We hope our proposed hybrid model can serve as an effective tool to forecast the price of carbon credits.