This research presents a proposed method to overcome difficulties which we face when we use traditional regression analysis method and provide solutions in an automated manner that does not require experience or a large field study. Some of these difficulties are mathematical difficulties, some of which are the need for considerable experience to choose the appropriate data model, especially when the traditional regression analysis method is not available. The new approach raises forecast accuracy with powerful Neural Network MATLAB code. This approach automatically trains neural networks and applies them for regression analysis, thus getting accurate, business or stock market predictions doesn't require much effort or time. This means faster and more precise results than ever before.