Abstract: Embodiment of the present disclosure discloses method and system for predicting a success rate of a project. The method comprises initiating a natural language conversation with stakeholder associated with project to determine one or more features relevant for the stakeholder. The one or more features are determined by performing Bayesian network analysis for one or more properties associated with project. The method comprises creating relationship structure, comprising one or more features, based on natural language processing of natural language conversation, wherein each of one or more features are assigned a score. The method comprises creating a prediction model with relative weightage values to each of one or more relevant features based on assigned score, wherein the prediction model is trained based on historic data associated with project and predicting success rate of project based on trained prediction model and current state of the project.