Is a PhD worth it in machine learning?

I'm in a graduate program in statistics right now, and have dabbled in machine learning and data mining for a while. I'm currently getting my master's, and I'm trying to decide if it's worth it to get a Ph.D if this is the field I want to get into.

What is your experience with the MS/PhD requirements for ML/DM jobs? Is a master's or master's + experience equivalent to a phd for most jobs? Most research jobs? And are most ML/DM jobs research oriented or not?

Is having a degree in statistics, instead of CS, a hindrance? Can a strong statistical background make up for a less strong CS background?

Lastly, what skills are most valuable for work in machine learning, or what skills are too often overlooked?