Predicting the company's future revenue accurately is the most important thing for rational investors. In this paper, a hybrid prediction model based on multi-factor features and time series features is proposed to predict company revenue by using model fusion. The linear regression model and the XGBoost model are severed as base models for a fused model. First, we conduct analysis and research based on the collected historical data of listed companies. In addition, these explicit and implicit factors are selected as the important features to feed into the model. The classical regression model was used to compare with the fusion framework proposed in the paper to prove high precision and superior performance of the later. Finally, our data-driven experiments show that the company's revenue forecast accuracy has got a great improvement and has strong interpretability.
[1]
Mohsen Akbari,et al.
Financial forecasting using ANFIS networks with Quantum-behaved Particle Swarm Optimization
,
2014,
Expert Syst. Appl..
[2]
Lawrence D. Brown,et al.
Comparing the Value Relevance of Two Operating Income Measures
,
2001
.
[3]
J. Friedman.
Greedy function approximation: A gradient boosting machine.
,
2001
.
[4]
Tianqi Chen,et al.
XGBoost: A Scalable Tree Boosting System
,
2016,
KDD.
[5]
Chen Ping,et al.
Forecast of Yearly Stock Returns Based on Adaboost Integration Algorithm
,
2017,
2017 IEEE International Conference on Smart Cloud (SmartCloud).