Markov chain simulation code

In Markov chain code

transition_probabilities = [0.1 0.9;0.8 0.2];

starting_value = 1;

chain_length = 5;

chain = zeros(1,chain_length);

chain(1)=starting_value;

for i=2:chain_length

this_step_distribution = transition_probabilities(chain(i-1),:);

cumulative_distribution = cumsum(this_step_distribution);

r = rand()

chain(i) = find(cumulative_distribution>r,1);

end

whats the aim of using cumsum function?

Why compare with random value in each iteration?