An example of some time series input I have in mind. Suppose I have multiple users working on a 20-minute search task, using a commercial search engine like Google, Yahoo, or Bing. It's the same task, but users may finish after 10 minutes if they feel they have completed the task to the best of their ability. The y-axis is a "performance metric" (like normalized discounted cumulative gain), and measures are taken either:
- At each query
- At each minute
Both the # of queries and time are obviously not constant across users. The goal here is for visualization/visual comparison, but if there's methods that are more suited to recommendation/prediction, I'm happy to hear those solutions, too!