Function 如何使用Joblib或其他并行计算优化这个python函数？

使用此函数计算距离矩阵需要花费大量时间。如何优化它？我应该使用作业库/并行计算吗？

这是在一个ML项目的服务器上运行的，执行此任务/功能需要很多时间

寻找优化Python代码的技巧

def get_dtw_matrix(self, lotrunnums, target_lotrunnums=None, data_targets=None, data_clusters=None, data_train=None):
    now = datetime.datetime.now()
    #print("inside get_dtw_matrix function - AMKPIV_SP.py",now.strftime("%Y-%m-%d %H:%M:%S"))
    if data_train is None:
        data_train, metric_preprocess = self.preprocess()        

    data_lotrunnums=self.get_data_to_dict(data_train, lotrunnums)
    #print("data_lotrunnums calculated",now.strftime("%Y-%m-%d %H:%M:%S"))
    if data_targets is None:
        #print("data_targets is None here...",now.strftime("%Y-%m-%d %H:%M:%S"))
        data_targets=self.get_data_to_dict(data_train, target_lotrunnums)
    
    data=data_lotrunnums
    targets=[]
    clusters=[]
    targets=data_targets.keys()
    data.update(data_targets)        
    if data_clusters is not None:
        clusters=data_clusters.keys()
        data.update(data_clusters)
    
    lotrunnums=data.keys()
    indexes, columns=lotrunnums, lotrunnums
    distance_matrix = pd.DataFrame(index=indexes, columns=columns)    
    lotrunnums=distance_matrix.columns
    for i in range(len(lotrunnums)):
        lotrunnum1=lotrunnums[i]
        trace1=data[lotrunnum1]
        for j in range(i, len(lotrunnums)):
            lotrunnum2=lotrunnums[j]
            trace2=data[lotrunnum2]
            if lotrunnum1==lotrunnum2:
                distance_matrix.loc[lotrunnum1, lotrunnum2]=0
            else:
                distanceDTW=DTWDistanceAMKPIV(self.distance_tol, self.distance_lagmax,self.distancetype, self.remove_repeated)
                res1=distanceDTW.distance_to_target(trace2, trace1)
                if lotrunnum1 in targets:
                    distance=float(res1[0]["distance"])
                else:
                    res2=distanceDTW.distance_to_target(trace1, trace2)
                    distance=min(float(res1[0]["distance"]), float(res2[0]["distance"]))
                distance_matrix.loc[lotrunnum1, lotrunnum2]=distance
                distance_matrix.loc[lotrunnum2, lotrunnum1]=distance    
    indexes=dict({"target": list(targets), "cluster": list(clusters)})
    #print("returning data of get_dtw_matrix",now.strftime("%Y-%m-%d %H:%M:%S"))
    return distance_matrix, indexes