I have a df which consists of two columns, 'date' and 'value', containing multiple years data of 'value.' I'd like to create a boxplot of the range of values grouped by month and year. I use groupby to group the values as such:
df.groupby(df['date'].dt.strftime('%b%Y'))['value']
But now I cannot figure out how to operate on the resulting groupby data structure to create the boxplot (via either matplotlib or seaborn). Stumped.
Let's assume your dataframe looks like it was generated by the following code:
import pandas as pd
import numpy as np
df = pd.DataFrame({'date': pd.date_range('2019-1-1', periods=1000, freq='D'),
'value': np.random.randint(0,10,size=1000)})
It looks like this:
date value
0 2019-01-01 3
1 2019-01-02 9
2 2019-01-03 2
3 2019-01-04 2
4 2019-01-05 4
.. ... ...
995 2021-09-22 3
996 2021-09-23 5
997 2021-09-24 1
998 2021-09-25 4
999 2021-09-26 8
Now you can compute a new column month:
df['month'] = df.date.dt.month
date value month
0 2019-01-01 3 1
1 2019-01-02 9 1
2 2019-01-03 2 1
3 2019-01-04 2 1
4 2019-01-05 4 1
.. ... ... ...
995 2021-09-22 3 9
996 2021-09-23 5 9
997 2021-09-24 1 9
998 2021-09-25 4 9
999 2021-09-26 8 9
And call boxplot using this new column as grouper:
df.boxplot('value', by='month')
parse_dates your time series Date column and set it as the index. Use pd.Grouper and level 0 to access the index and set the frequency to D or daily. The boxplot will find the min and max frequency and median and outliers based on the grouped.index.year
df=pd.read_csv('sp500.csv',parse_dates=['Date'],index_col='Date')
grouped=df.groupby(pd.Grouper(level=0,freq='D'))['Open'].mean()
sns.boxplot(x=grouped.index.year, y=grouped.values,
data=grouped)
plt.xticks(rotation=90)
plt.show()
You can groupby using the datetime accessor to_period() with a monthly frequency.
import pandas as pd
import numpy as np
df = pd.DataFrame({'date': pd.date_range('2019-1-1', periods=1000, freq='D'),
'value': np.random.randint(0,10,size=1000)})
df.groupby(df.date.dt.to_period("M")).boxplot(subplots=False, rot=90)
