I have built a kNN model using scikit learn that is able to predict a binary outcome very well. The data itself is quite basic, it is simply a 1-D waveform. When feeding the waveform into the model it predicts. However, I would like the data to be streamed through the model if that makes sense, but I am running into issues how to implement.
My first thought was to use a sliding window, but I don't know if that's the best way to go about things because depending on the speed at which data arrives, what if I have a case where there are two distinct peaks in my waveform that require two separate classifications. In that case, if I make my window small, then there is not enough data to make a proper classification. Any guidance or literature on methods would be great.