时间序列趋势判断（一）——斜率阈值判断

用最小二乘法拟合 ax+b 的a值，代表斜率，

• 如果abs(a) > 0.1763 (这是tan10度，相当于一个阈值，如果拟合的曲线仰角超过10度，就证明有倾向)

示例代码

import numpy as np


def trendline(data):
    """拟合后用斜率判断
    """
    index = [i for i in range(1, len(data) + 1)]
    coeffs = np.polyfit(index, list(data), 1)
    slope = coeffs[-2]
    if slope > 0.1763:  # 0.1763是 tan10°
        return "increasing"
    elif slope < -0.1763:
        return "decreasing"
    else:
        return "no trend"


if __name__ == '__main__':
    data = np.array([1, 2, 3, 4, 3, 5, 5, 7, 4, 7, ])
    print(trendline(data))