Not sure what precision level this problem is looking for, but for right-skewed distributions, we know that the mean is going to be pulled right and therefore the mean should be larger than the median. To a high confidence level, the mean should fall between 50 and 59, or in the third column.
If a single estimation is wanted, assume the values inside each column are uniformly distributed:
150 is the amount of interest earned for one year because the formula of PRT is, in this case 5000 times 0.03 times one which gets you the answer of 150.