while 1:
ini_r, ini_alpha, ini_a, ini_b = np.random.normal(1, 0.05, 4)
cost_function = CostFunction(sum_data)
optimizer = Optimizer(sum_data,
np.array([ini_r, ini_alpha, ini_a, ini_b]),
norm_factor)
params_gen, score_gen = optimizer.genetics(
cost_function.neg_log_likelihood, sum_data,
np.array([ini_r, ini_alpha, ini_a, ini_b]), norm_factor)
params, score = optimizer.nelder_mead(
cost_function.neg_log_likelihood, sum_data,
np.array([ini_r, ini_alpha, ini_a, ini_b]), norm_factor)
if not is_gen_preffered and score_gen < 4.515:
params_final.append(params)
scores.append(score)
break
if is_gen_preffered and score < 4.515:
params_final_gen.append(params_gen)
scores_gen.append(score_gen)
break
if not is_gen_preffered:
df = pd.DataFrame(params_final, columns=['r', 'alpha', 'a', 'b'])
df['scores'] = scores
df.to_csv('input/estimated_parameters.csv')
