This paper proposes a hosting capacity analysis and forecasting model for power distribution networks based on machine learning algorithms. The forecasting model is built using a variety of machine learning techniques, including multiple linear regression (MLR), multivariate linear regression (MVLR), and support vector machine (SVM). Pearson's correlation coefficient is used to select input features from a set of input variables. For estimating the hosting capacity of distributed energy resources (DERs), this study uses the IEEE 13 bus network as a test system. The results show that the MVLR provides better performance than other comparable models in terms of very low mean absolute percentage error (0.15%), root mean square error (12.93), and 97% accuracy for hosting capacity prediction. The proposed approach allows grid operators to successfully manage the integration of DERs into the electricity distribution system by accurately estimating and forecasting hosting capacity.