Predicting project sales prices using machine learning techniques: a case study in a project consultancy

. The article aims to apply a comparative analysis of machine learning techniques to predict project sales prices of a consulting company located in Curitiba/PR. The company has projects in the most diverse fields, such as in the strategic, productive, quality and innovation areas. Due to this diversity, company managers find it difficult to calculate the sale value of new projects, since they deal with different types of predictor variables such as: type of consultant, type of project and number of hours. In this sense, there is a need to use a method that predicts from a multivariate analysis and that results in sales values close to those expected by the company. To this end, a literature review was carried out on the research topics, namely: Production Planning and Control (PPC) and machine learning techniques; then, the company's current sales prospecting process was mapped; in addition, data were collected, analyzed and prepared, and then proceeded to the testing stage and selection of the best model; and finally, the improvement proposal was discussed with the organization. As a result, it was obtained that the application of the Gradient Boosting Machine (GBM) technique obtained the lowest error result among the Machine Learning techniques tested. The error was approximately 21%, which can be considered acceptable for the analyzed segment. Thus, this work met the expectations of stakeholders by presenting the possibility of pricing projects using computational algorithms for forecasting demand.