you have 3 sets of data of 20 sample points each.
each set represents 1 of the 3 days.
the data tells you the following:
day sunny rain total p(rain)
1 11 9 20 9/20
2 9 11 20 11/20
3 9 11 20 11/20
the experimental probability that it will rain on all 3 days is 9/20 * 11/20 * 11/20 = .136125
the experimental probability that it will not rain on at least 1 of those days = 1 minus the probability that it will rain on all 3 of those days which is equal to 1 - .136125 = .863875.
my interpretation of this problem is that each day had 20 coin flips to determine the probability that it would be sunny on that day or rainy on that day.
the results were based on that interpretation.