LSTM Classification

Guys how does increasing or decreasing the number of timesteps affect my LSTM model…does it affect my accuracy or time taken to train?
…I have a timeseries dataset where I have to classify it based on 2 parameters into 3 classes

Hi aravinda,

Firstly I am not the expert, but I think increasing or decreasing timesteps depends your future predicts. For example, you want to forecast the weather for the next week or you just want 3 days.

Let’s say you have 10000 training sets and the accuracy rate is sufficient for this 3-day forecast, but the accuracy may not be sufficient for 1 week. Therefore, by enlarging the training set a little bit (by adding more historical data), you can reach the desired accuracy rate. But this will increase the training time.

anim_sound3