When the problem says theoretical probability of a fair cube, you assume that every outcome is equally likely.
The theoretical probability of rolling a number less than 4 with a fair, 6 sided dice is 50% or 3/6=1/2. This is because you can roll either a 1, 2, 3, 4, 5 or 6. half of those outcomes qualify as "less than four".
To get the experimental result you have to do some more arithmetic.
7+12+10=29. There are 29, out of 50 trials, where the dice landed on less than 4. so the experimental results are that 29/50=58% of dice rolls are less than 4.
So the difference between the theoretical probability and the observed, experimental results is 58-50=8%