The earthmoving operation has significant importance in building projects, constituting around 30% of the total cost. Numerous research studies have put forth diverse methodologies for evaluating earthmoving performance to mitigate the potential inaccuracies in estimating the costs and duration of earthmoving activities. In the phase of detailed estimation, the project scope and limitations regarding project information may present obstacles to planners in their efforts to estimate project costs and duration, particularly in the context of fleet selection. However, a significant number of studies fail to adequately consider the importance of budget and schedule planning during the detail estimating phase before the commencement of construction. This oversight occurs despite the uncertain working conditions and restricted availability of data that often characterize this period. In the conventional approach, the attainment of optimal equipment selection has often been facilitated via the use of the construction standard productivity rate (CSPR) or equipment manuals, alongside the discrete event simulation (DES) methodology. Nevertheless, the estimation of system productivity using CSPR is characterized by inefficiency, while planners lacking simulation knowledge may encounter difficulties in utilizing DES. Consequently, the present study has duly considered these factors by incorporating a machine learning (ML) predictive model for earthmoving productivity. This model encompasses the earthwork process and utilizes synthetic data generated from the discrete event simulation (DES), with inputs derived from the construction site production rate (CSPR). Additionally, the study integrates multi-objective optimization algorithms to assist planners in selecting the most optimal fleet combination. Prior to selecting a suitable machine learning (ML) model, several methodologies are explored and evaluated in order to integrate them into the NSGA-II (Non-dominated Sorting Genetic Algorithm II) multi-objective optimization algorithm. This paper presents a case study analysis of a genuine project, serving as an illustrative example of the practical use of the proposed technique in a real-world planning scenario. The research aims to provide effective assistance to construction planners throughout the detail estimating and bidding phases by accurately recommending the most suitable combinations of equipment based on several earthmoving factors such as road type, soil conditions, and hauling distance. Furthermore, it is both practical and advantageous for planners to engage in strategies aimed at addressing the challenges posed by unclear project scopes and recurrent simulation tasks in the estimating process. Finally, this study also includes a comment on the limits and potential areas for future development.