tensor_run[0, run, t_p] = tp_a[run * 1 + 0]
tensor_run[1, run, f_p] = fp_b[run * 1 + 0]
tensor_run[1, run, f_n] = fn_b[run * 1 + 0]
tensor_run[1, run, t_p] = tp_b[run * 1 + 0]
# print tensor_run
new_n = 2
new_tensor_run = tensor_run
#new_n=3
#new_tensor_run = tensor_run[[a,a,b],:,:]
total_f, col_seq = select_by_round_robin(
new_tensor_run,
np.ones(new_n, dtype=int) * -1, [], [], new_n * 2, True)
best_sum_total_f[
d], best_col_seq[d] = select_by_estimated_max_f_impact(
new_tensor_run, np.ones(new_n, dtype=int), total_f,
col_seq, estimated_scores, future_steps,
True) # Flight = 9, Blackoak 7, Hospital=5
#print best_col_seq
average_best = np.sum(best_sum_total_f.values(), axis=0) / float(
len(best_sum_total_f))
#print "a: " + str(error_fractions[a]) + " b: " + str(error_fractions[b]) + " -> " + str(list(average_best))
print str(list(average_best))
for run in range(runs):
for col in range(n):
tensor_run[col, run, f_p] = fp[col + n * run]
tensor_run[col, run, f_n] = fn[col + n * run]
tensor_run[col, run, t_p] = tp[col + n * run]
# print tensor_run
new_n=2
new_tensor_run = tensor_run[[a,b],:,:]
#new_n=3
#new_tensor_run = tensor_run[[a,a,b],:,:]
best_sum_total_f[d], best_col_seq[d] = select_by_round_robin(new_tensor_run, np.ones(new_n, dtype=int) * -1, [], [], future_steps, True) #Flight = 9, Blackoak 7, Hospital=5
#print best_col_seq
average_best = np.sum(best_sum_total_f.values(), axis=0) / float(len(best_sum_total_f))
#print "a: " + str(error_fractions[a]) + " b: " + str(error_fractions[b]) + " -> " + str(list(average_best))
print str(list(average_best))
'''
labels = []
start = 0
for ii in range(new_n):
start +=4
labels.append(start)