Hi,
I am working on Stock Market Prediction using Artificial Neural Networks. I am using MATLAB software .To get started in understanding the software, i have just used daily data of S&P BSE Auto, a sectoral index traded at Bombay Stock Exchange of India for a period of 12 years. I have collected the Open, High, Low and Close data of BSE Auto and runned the model. Herewith, I have attached the output.
Kindly help me in getting inference for the output and i require clarification whether the analysis done is right or wrong.
I seek your help in my research endeavours.
THANK YOU VERY MUCH.
hi mr kumar
you can use mse commnd or plot confusion matrix in matlab
please read neural network manual in matlab. it is so useful.
As predictions calculated by ANN is machine learning, the complete set of data is divided into 3 groups. Training set is used for teching the model. Validation is used to determain when to stop learning (protect from overfitting). Test dataset is used to calculate the errors of the model built but on the data that was not seen by the machine before. Linear correlation between predicted values and the real values can be calculeted for each data set separately or fore whole set. I haven't done any prediction with time series, but the architecture of ANN is clear: you can define activation function separately for hidden layer(s) and for output layer. The same with neurons in the hidden layer. Delays belong to time series (not my domain).
Best regards
Hubert
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Software