数据预处理：PySpark 的实现线性插值填充缺失值

数据预处理：PySpark 的实现线性插值填充缺失值

1. Python 实现线性插值填充缺失值

实现函数为：

def linear_insert(x1, y1, x2, y2, insert_x):
    if type(insert_x) == int:
        insert_x = [insert_x]
    k = (y2 - y1) / (x2 - x1)
    return [k * (x - x1) + y1 for x in insert_x]


def fill_na_by_linear(lst):
    first_flag = False
    first_na = 0
    length = len(lst)
    for i in range(length):
        item = lst[i]
        if not first_flag:
            if item is None:
                first_na = i
                if first_na == 0:
                    # 第一个缺失值填充为 0
                    lst[0] = 0.0
                    continue
                first_flag = True
        else:
            if item is not None:
                first_flag = False
                lst[first_na:i] = linear_insert(first_na - 1, lst[first_na - 1], i, lst[i], ran