%posterior predictive - 
% P(X_n+1) 
for i=1:MM
    post_pred_sample(i,:) = expected_function(posterior_sim(i,1),posterior_sim(i,2),posterior_sim(i,3));
%    post_var_sample(i,:) = variances_func(posterior_sim(i,1),posterior_sim(i,2),posterior_sim(i,3));
end
%now have all the predicted values of fails per month for months 1:m+m2
%now pick out the one that is of interest.
%am assuming that this is called from outline 2 which means that we are
%interested in predicting month count+1 (outline2 uses the first "count"
%observations to estimate the parameters.
sorted_post_pred(:,(count+1)) = sort(post_pred_sample(:,(count+1)));
upper(count+1)=sorted_post_pred(int32(MM*.025),(count+1));
lower(count+1)=sorted_post_pred(int32(MM*.975),(count+1));

%post_v = mean(post_var_sample);
post_prediction(count+1) = mean(post_pred_sample(:,(count+1)));
%post_var(count+1) = post_v(count+1)
display('#failed in month I+1 -- #predicted -- 95%upper -- 95%lower')
[obs(1,(count+1)) post_prediction(count+1) upper(count+1) lower(count+1)]