Predicting Financial Situation for Companies Through Integration of Adaboost Algorithm and BP Neural Network

In study, a prediction model is proposed based on combined Adaboost algorithm and BP neural network in order to predict company’s financial situation and improve prediction accuracy of BP neural network model. The significance of solving the problem is to adjust the company's financial expenditure and make better forecast and analysis for the development of companies. The proposed method regards BP neural network model as weak predictors and uses Adaboost algorithm to construct strong predictor, which solves the problems of local minima defects and slow convergence of BP neural network model. The core innovation is to construct strong predictor using Adaboost algorithm in the research. The efficiency of the proposed prediction model is proved by training and predicting 1350 groups of statistical data of company’s financial situation. The computer simulations have shown that the model is effective and suitable, has higher forecasting accuracy and is applicable to practice compared with previous work.