For machine learning classification or regression, such as adaboost or SVM.
In the case of live trading, is the output also required or is only the input required?
Your question is not very clear. Can you add examples/ explanation? Usually the inputs to a ML model is price (or transformed price) data. The output is a classification or regression prediction.
Hello Prodipta!
My question was that when I train and test (cross-validation) the model, I have to give the input and the output for the model.
What about deployment?
When the model is saved and ready for deployment, is it still necessary to gave the input and output to the model?
Should the input and output data be feed to the model even in the case of deployment?
You need output only when you need to train in supervised learning. For prediction, you supply the inputs and the model gives you the output (the model prediction). This is the whole point of training - the model learns the function that maps input to output during training, and use that learnt function to generate output during prediction.
Sometimes models are trained "online", that is, it continue to train incrementally. In such cases, since the traning continues, you will still need to collect the realized output and feed it back to the model during the deployment.
I suggest you complete some basic course on machine learning before you proceed further.