# Time Series Graphics

# Seasonal periods

Some graphics and some models will use the seasonal period of the data. It is the number of observations before the seasonal pattern repeats
For quarterly, monthly and weekly data, there is only one seasonal period — the number of observations within each year.
If the data is observed more than once per week, then there is often more than one seasonal pattern in the data.

from pyconfig.utils import *
import statsmodels.datasets as smd
import matplotlib.pyplot as plt
import darts.timeseries as ts

df = smd.get_rdataset('melsyd', 'fpp2').data

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 283 entries, 0 to 282
Data columns (total 3 columns):
 #   Column          Non-Null Count  Dtype  
---  ------          --------------  -----  
 0   First.Class     282 non-null    float64
 1   Business.Class  176 non-null    float64
 2   Economy.Class   282 non-null    float64
dtypes: float64(3)
memory usage: 6.8 KB

df.shape

(283, 3)

df.head()

	First.Class	Business.Class	Economy.Class
0	1.912	NaN	20.167
1	1.848	NaN	20.161
2	1.856	NaN	19.993
3	2.142	NaN	20.986
4	2.118	NaN	20.497

plt.plot(df)
plt.legend(['First Class', "Business Class", "Economy Class"])
plt.show()