Abstarct: One of the concerns of software developers is always the issue of testing the software project and finding software errors that may occur in the project. But this process is a time consuming activity with a lot of time and resources. The concept of software fault prediction is a model that can greatly contribute to the concept of software testing using a limited amount of resources and time. Predicting software errors is in defining a process for predicting potential errors in software modules and finding accident hotspots in the modules, which we do using the prediction model we have built. A common predictive model is built with machine learning and techniques, the sources of which in construction and training can be a set of error data that includes a large number of software features. Now with this forecasting model we can examine new projects and predict the possibility of errors in different parts of it. One of the types of software error prediction models is the number of software error prediction model. This model estimates the number of errors that can occur at the level for which it is defined. These levels can include classes, modules, functions, etc. In this research and according to the subject under study, the level of functions is located.