I want to implement a statistical approach that can detect change points in real-time streaming data. Specifically, for every newly acquired data point, I want to evaluate whether it represents a change point or not.
I used to use financial transactions to predict the stock market, similar to what you’re trying to do. My model wasn’t perfect, but it performed better than guessing randomly. I used something like LSTM (Long Short-Term Memory) for this. However, you need to provide more information about your data. What kinds of changes are you expecting to see in your data?
I want a statistical approach to detect online change point