Hi, trying to figure out ml in opensearch… is it possible to evaluate models? I mean, after training phase, can you investigate how well the model fit to non viewed data? is there an API for this?
Sorry that missed your question. We don’t have dedicated API to evaluate models now. Can you elaborate more about your use case? For example, how will you split your training and test data, will the data be labeled or not, what criteria/result you care and want to see in the evaluation report (for example, the precision/recall/model latency etc), would you prefer to compare two/multiple model’s evaluation result? More details can help use understand more about your use case and build a better feature.