What is the term when I use multiple models and when in each prediction I use the one that performs the best, and where I can read on it?

Hello,
Like in topic.
I know there is the term for that but I have it forgotten.
To make it more precise - I have now prepared my whole dataset with various "indicators" calculated using raw data.
Now I want to assemble around 20 models where each model will proccess raw data and few of these indicators, and I want these models to "co-operate" so the prediction which I will be using later will be the best prediction from each …

neuralnetworks prediction