the best way to consider probability problems such as these is by using the definition of probability and adjusting that to reflect the empirical nature of the observations. In fact such probabilities are called empirical probabilities.
so by definition the probability of an event, E, is given by
where n(E) is the number of ways the event can occur and n(S) is the total number of outcomes in the sample space. For an empirical probability this can be adjusted by letting n(E) be the number of times an event was observed; similary let n(S) be the number of trials conducted.
so for your first example:
E is the event that an odd sum was showing
n(E) it the number of times an odd sum was observed, 42
and n(S) is the number of trials, 100. so:
P(E)=42/100 or .42
i'm confident you can finish from there.