Selected Answer
Presume that you received 43 "good" cases and 5 "bad" cases", then 5 would be 10.42% of the total number of cases received (43 + 5 = 48), meaning 10.42% of all cases received were bad. This you correctly calculate using the formula =G2/SUM(F2:G2). The formula =F2/SUM(F2:G2) would return 89.58%, meaning that 89.58% of all cases received were good. The good + bad = 100% = 48 cases.
Note that each row is self-contained with no data from any other row used. This also goes for the last row where, in fact, 22.41% is the average percentage of "bad" cases received. The average is calculated by merging the quantities of cases. It can't be calculated by applying the average of percentages calculated from, essentially, unrelated numbers.
Perhaps this formula is what you are after.
[H2] =G2/SUM($F$2:$G$9)
Here the 5 "bad" cases in G2 are calculated against the total number of received cases (90 + 26 = 116) and the results in H2:H9 would add up to 22.41%