Function cal_max_c_alignment

union/TUS/run_c_alignment.py:193–220

(model, folder_path = 'benchmark')

def cal_max_c_alignment(model, folder_path = 'benchmark'):
    file_names = os.listdir(folder_path)
    #找出表中最大的列数
    max_c = 0
    for file_name in file_names:
        df = pd.read_csv(folder_path+'/'+file_name)
        max_c = max(max_c, len(df.columns))
    #使用字典存每个max_c的分布，key是c，value是列表
    max_c_score_dict = {}
    for i in range(max_c):
        max_c_score_dict[i+1] = []

    np.random.seed(41)
    combinations = list(itertools.combinations(file_names, 2))
    random_numbers = np.random.randint(low=0, high=len(combinations), size=800)#11696
    subset = [combinations[i] for i in random_numbers]
    idx = 0
    for pair in subset:
        d = alignment_process(pair[0], pair[1], model, 'distribution')
        for c, score in d.items():
            max_c_score_dict[c].append(score)
        idx += 1
        print(idx)

    for c, l in max_c_score_dict.items():
        file_path = 'alignDistribution/'+str(c)+'_align_distribution.json'
      
        save_dict_to_json(l, file_path)

#首先读取所有的表，循环处理，每次处理一个表
def main():