Shouldn't P(increase | happy) = P(increase) P(happy | increase) / P(increase) be P(increase | happy) = P(increase) P(happy | increase) / P(happy)? Also I am no

Shouldn't P(increase | happy) = P(increase) P(happy | increase) / P(increase) be P(increase | happy) = P(increase) P(happy | increase) / P(happy)? Also I am not getting why he cannot simply compute 950/1000 = 95%? How does Bayes rule help us in his example? What am I missing?