People often talk about skills needed to be a good data scientist. Maybe general, like python or R, or specific, like "split-apply-combine", "linear algebra", "ggplot".
By far the most useful is the desire to conquer error messages or crazy results, no matter how long it takes