基于fbprophet框架的期末余额预测方法

Ending Balance Forecasting Method Based on fbprophet Framework

时间序列分析充分揭示了动态数据的变化规律,因此在金融、会计等需要体现时间特征的领域得到了广泛应用。本文使用Facebook开源的时间序列预测框架fbprophet,针对单位或企业的期末余额预测问题进行了建模和试验。期末余额指某一时间段内期末结出的账户余额。期末余额预测有助于单位或企业更好的了解余额走向,活用库存资金,提高交易收益。我们以天为单位,预测未来一个月内的期末余额。通过和MLP、XGBoost等深度学习、机器学习模型预测结果的比较,fbprophet的MAPE表现最优。实验结果说明相对于其他算法,fbprophet在情况复杂、规律性差的数据集上仍可得出较为准确的预测结果。

Time series analysis fully reveals the changing law of dynamic data, so it has been widely used in fields such as finance and accounting that need to reflect time characteristics. This paper uses Facebook’s open source time series forecasting framework, fbprophet, to model and test the ending balance forecast problem of a unit or enterprise. The ending balance refers to the balance of the account at the end of a certain period of time. The ending balance forecast helps the unit or enterprise to better understand the balance trend, use inventory funds, and increase trading income. We forecast the ending balance in days for the next month. Compared with the predicted results of deep learning and machine learning models such as MLP and XGBoost, the performance of MAPE of fbprophet is the best. The experimental results show that compared with other algorithms, fbprophet can still obtain relatively accurate prediction results on data sets with complex conditions and poor regularity.