lasso.py 文件源码

def transcoding(x):
    l1 = list(x)
    province = list(set(l1))
    n = len(province)
    mat = [[0 for j in range(n)] for i in range(n)]
    province_dict = {}
    for i in range(n):
        mat[i][i] = 1
        province_dict[str(province[i])] = mat[i]
    ret = []
    for i in range(len(l1)):
        key = str(l1[i])
        ret.append(province_dict[key])
    return pd.DataFrame(ret),province_dict


#hot_coding,province_dict = transcoding(df[10])