| Summary: | 碩士 === 國立彰化師範大學 === 商業教育學系 === 92 === Abstract
A Study of EPS Forecasting by BPN— Media Tek Incorporation Case
Financial statements reflect performance of corporations. Management, investors and researchers use the information to reach diagnosis, evaluation and prediction of earnings. Therefore, it is imperative for management and investors to have a model capable of forecasting earnings correctly. This study intends to apply BPN to the prediction of earnings. We use financial statements from 30 electronic tech stocks and run the information through BPN to create a model. Then, we use the model to make earnings prediction for Media Tek Incorporation. We found:
1. BPN can be an effective model for earnings prediction.
2. Variables in this model include current assets, net long-term equity investments, intangible assets, other assets, operating revenue, operating cost, operating expense, and net non-operating revenue expense.
3. BPN model can make correct prediction about 93% of the time. Therefore, it is a viable model for earnings prediction.
4. Of all the variables, net long-term equity investments, other assets, operating revenue, operating cost, operating expense and net non-operating revenue expense have significant correlation to earnings prediction.
In conclusion, we find the BPN is an excellent model for earnings prediction.
Keywords：BPN，Artificial Neural Networks，
EPS Forecasting， Forecasting
|
|---|
