
Senior advised me just to watch the alerts from the server. I'm Data Analyst intern, I don't care about pings on the server. I looked through the metric graphs, and noticed the “Predictive Forecasting” settings. The "Training Window" slider (30 days, 90 days, or customized offset) can be adjusted manually to train this Holt-Winters math model. This is possible even for changing confidence interval! If you're not one of the happy normal users, but you're a math nerd, you can see the HTML5 canvas training forecast line change when you change the training offset (if you don't, you can lock it), and that's pure heaven. Review collected by and hosted on G2.com.
This is a beautiful feature that they don't publicise! The training slider is deep within the two sub-menus in widget settings. Other than that, if I export to CSV, it only includes historical raw data and not forecasted coordinate points computed. I don't like having to copy coordinates for my Python script! Review collected by and hosted on G2.com.