Compute average case with best case

is it correct to compute the average case time complexity of an algorithm by taking the mean of the best and worst cases ? My findings : for binary search, $frac{log (n) +1}{2}in Theta left(log (n )right)implies $ the average case is $log (n).$