By simply stopping and restarting the datafeed of the ML job at the appropriate times, you can create a gap in the analysis. Here, the datafeed was stopped at midnight on February 9^th and restarted at midnight on February 10^th:

It was like February 9^th never happened!