python 时间序列分解 stl_jupyter notebook运行时间序列分解STL趋势和季节

1. "ImportError: No module named" when trying to run Python script

notebook不知道module的路径在哪，默认只知道current path

先找到包所在位置

pip show --verbose pandas

再告诉notebook

import sys

sys.path.append('/usr/local/lib/python2.7/dist-packages')

2.  安装分解时间序列用到的包

pip install statsmodels

3. AttributeError: 'Index' object has no attribute 'inferred_freq'

seasonal_decompose()对pandas Series需要是datetime

import pandas as pd

from pandas import Series

from matplotlib import pyplot

from statsmodels.tsa.seasonal import seasonal_decompose

df = pd.read_csv('data.csv', header=0)

df.Date = pd.to_datetime(df.Date,errors='coerce')

df = df.set_index('Date')

result = seasonal_decompose(df, model='mutiplicative',freq=7)

result.plot()

pyplot.show()

4. notebook显示图片太小，变大

import matplotlib.pyplot as plt

fig_size = [20, 9]

plt.rcParams["figure.figsize"] = fig_size