[Python] Visualize the heat of Tokyo and XX prefectures (DataFrame usage memo)
0. Problems hidden in everyday life
Summer greets in the state of emergency. In order to avoid the Three Cs thoroughly, let alone play, I usually go through ideas I couldn't go to the cafe I was given, so I kept it at home. The only topics are the number of people infected with the coronavirus and the heat.
As a person living in Tokyo, even if I talk about how hot it is and I can't move, my cousins and father told me that the local area is just as hot and that I am doing this kind of exercise (cycling) in the hot weather. It seemed that he was competing for the heat, and he was tired.
In the news that I see from time to time, it seems that the locals in the snowy country are hitting the highest temperature as in Tokyo.
It was so hot that I thought it was steaming from the ground. Just walking for 10 minutes makes me flutter. And even if you go into a cafe and have a cold drink, you can't easily get rid of the heat that fills your body. While having a drink, I thought about how to compare the heat of Niigata and Tokyo at the highest temperature.
1. 1. Acquisition of temperature data
From Japan Meteorological Agency, we were able to obtain weather information for each region by following the procedure below. This time we need data from Niigata and Tokyo, so even if the following is for Tokyo Execute and press the "Download csv file" button.
I edited only the item part of the above csv file to make it easier to use.
2. Various issues
(1) Read the CSV file
weather_parse2.jpynb
import pandas as pd
import datetime as dt
#Library to display graphs
import matplotlib.pyplot as plt
#Japanese localization library
import japanize_matplotlib
#Specify the character code and read
df = pd.read_csv("data_weather2.csv",encoding="sjis")
print(df)
__ Output result __
Date Average temperature(℃)_Tokyo Quality Information Homogeneous Number Maximum Temperature(℃)_Tokyo quality information.1 Homogeneous number.1 \
0 2020/1/1 5.5 8 1 10.2 8 1
1 2020/1/2 6.2 8 1 11.3 8 1
2 2020/1/3 6.1 8 1 12.0 8 1
3 2020/1/4 7.2 8 1 12.2 8 1
4 2020/1/5 5.4 8 1 10.2 8 1
.. ... ... ... ... ... ... ...
229 2020/8/17 30.8 8 1 36.5 8 1
230 2020/8/18 30.3 8 1 34.3 8 1
231 2020/8/19 29.0 8 1 34.2 8 1
232 2020/8/20 29.7 8 1 34.8 8 1
233 2020/8/21 30.1 8 1 36.0 8 1
Lowest Temperature(℃)_Tokyo quality information.2 Homogeneous number.2 Average temperature(℃)_Niigata Quality Information.3 homogeneous number.3 Maximum temperature(℃)_Niigata\
0 3.2 8 1 3.6 8 1 5.6
1 1.9 8 1 3.0 8 1 5.7
2 1.4 8 1 4.2 8 1 6.9
3 3.6 8 1 5.5 8 1 8.3
4 0.6 8 1 4.2 8 1 7.2
.. ... ... ... ... ... ... ...
229 27.2 8 1 27.8 8 1 30.4
230 27.8 8 1 26.3 8 1 29.0
231 25.6 8 1 26.8 8 1 30.9
232 25.8 8 1 28.5 8 1 33.7
233 26.0 8 1 27.5 8 1 30.8
quality information.4 Homogeneous number.4 Minimum temperature(℃)_Niigata Quality Information.5 homogeneous number.5
0 8 1 1.0 8 1
1 8 1 2.0 8 1
2 8 1 1.7 8 1
3 8 1 2.4 8 1
4 8 1 1.7 8 1
.. ... ... ... ... ...
229 8 1 26.3 8 1
230 8 1 23.3 8 1
231 8 1 22.0 8 1
232 8 1 22.8 8 1
233 8 1 23.8 8 1
[234 rows x 19 columns]
(2) Column selection and range selection to be graphed
__ (Memo 1) To select and display column name 1 and column name 2 from the above __
Specify DataFrame [['column name 1','column name 2']].
__ (Memo 2) Creating a DataFrame __
DataFrame name = pd.DataFrame (data, columns = list of column names you want to specify)
__ (Note 3) To convert column name 3 to the default date type (yyyy-mm-dd) __
DataFrame name ['column name 3'] = pd.to_datetime (DataFrame name ['column name 3'], format ='format before change of column name 3')
weather_parse2.jpynb
new_columns = ['date', 'Highest temperature(℃)_Tokyo', 'Highest temperature(℃)_Niigata'] #Column name
new_data=df[['date','Highest temperature(℃)_Tokyo','Highest temperature(℃)_Niigata']]
#recreate dataframe
new_df=pd.DataFrame(new_data,columns=new_columns)
#If you do nothing, it will not be recognized as a date type
print(new_df.dtypes)
#Convert date column to date type
new_df['date'] = pd.to_datetime(new_df['date'], format='%Y/%m/%d')
print(new_df.dtypes)
print(new_df)
__ Output result __
Date object
Highest temperature(℃)_Tokyo float64
Highest temperature(℃)_Niigata float64
dtype: object
Datetime64[ns]
Highest temperature(℃)_Tokyo float64
Highest temperature(℃)_Niigata float64
dtype: object
Date Maximum temperature(℃)_Tokyo highest temperature(℃)_Niigata
0 2020-01-01 10.2 5.6
1 2020-01-02 11.3 5.7
2 2020-01-03 12.0 6.9
3 2020-01-04 12.2 8.3
4 2020-01-05 10.2 7.2
.. ... ... ...
229 2020-08-17 36.5 30.4
230 2020-08-18 34.3 29.0
231 2020-08-19 34.2 30.9
232 2020-08-20 34.8 33.7
233 2020-08-21 36.0 30.8
[234 rows x 3 columns]
weather_parse2.jpynb
new_df2=new_df[(new_df['date'] >= dt.datetime(2020,8,1)) & (new_df['date'] < dt.datetime(2020,8,21))]
print(new_df2)
__ Output result __
Date Maximum temperature(℃)_Tokyo highest temperature(℃)_Niigata
213 2020-08-01 31.8 28.8
214 2020-08-02 31.5 30.3
215 2020-08-03 32.3 32.2
216 2020-08-04 33.1 32.7
217 2020-08-05 34.2 33.4
218 2020-08-06 33.1 33.6
219 2020-08-07 35.4 28.3
220 2020-08-08 32.3 24.8
221 2020-08-09 34.7 28.4
222 2020-08-10 35.2 35.6
223 2020-08-11 37.3 35.7
224 2020-08-12 35.8 31.4
225 2020-08-13 36.1 31.4
226 2020-08-14 34.2 32.0
227 2020-08-15 36.1 30.1
228 2020-08-16 35.4 31.4
229 2020-08-17 36.5 30.4
230 2020-08-18 34.3 29.0
231 2020-08-19 34.2 30.9
232 2020-08-20 34.8 33.7
- The date is displayed as yyyy-mm-dd.
(3) Graphing
weather_parse2.jpynb
#Specify the data you want to graph
new_columns3 = ['Highest temperature(℃)_Tokyo', 'Highest temperature(℃)_Niigata'] #Column name
new_data3=new_df2[['Highest temperature(℃)_Tokyo','Highest temperature(℃)_Niigata']]
new_df3=pd.DataFrame(new_data3,columns=new_columns3)
#graph display
new_df3.plot()
- Since the index value on the horizontal axis is the number of rows, I do not know the correspondence of temperature to the date.
- The graph is a line and it is difficult to compare
3. 3. Completed form (specify the index as a date and output the maximum temperature for the specified period)
weather_parse3.jpynb
import pandas as pd
import datetime as dt
#Read by specifying the character code, indexing the date
df = pd.read_csv("data_weather2.csv",encoding="sjis")
#Index by date
df.index=pd.to_datetime(df['date'], format='%Y/%m/%d').values
#Delete the original column
df = df.drop(columns='date')
print(df)
Output result
- The index is date type.
Average temperature(℃)_Tokyo Quality Information Homogeneous Number Maximum Temperature(℃)_Tokyo quality information.1 Homogeneous number.1 Minimum temperature(℃)_Tokyo\
2020-01-01 5.5 8 1 10.2 8 1 3.2
2020-01-02 6.2 8 1 11.3 8 1 1.9
2020-01-03 6.1 8 1 12.0 8 1 1.4
2020-01-04 7.2 8 1 12.2 8 1 3.6
2020-01-05 5.4 8 1 10.2 8 1 0.6
... ... ... ... ... ... ... ...
2020-08-17 30.8 8 1 36.5 8 1 27.2
2020-08-18 30.3 8 1 34.3 8 1 27.8
2020-08-19 29.0 8 1 34.2 8 1 25.6
2020-08-20 29.7 8 1 34.8 8 1 25.8
2020-08-21 30.1 8 1 36.0 8 1 26.0
quality information.2 Homogeneous number.2 Average temperature(℃)_Niigata Quality Information.3 homogeneous number.3 Maximum temperature(℃)_Niigata Quality Information.4 \
2020-01-01 8 1 3.6 8 1 5.6 8
2020-01-02 8 1 3.0 8 1 5.7 8
2020-01-03 8 1 4.2 8 1 6.9 8
2020-01-04 8 1 5.5 8 1 8.3 8
2020-01-05 8 1 4.2 8 1 7.2 8
... ... ... ... ... ... ... ...
2020-08-17 8 1 27.8 8 1 30.4 8
2020-08-18 8 1 26.3 8 1 29.0 8
2020-08-19 8 1 26.8 8 1 30.9 8
2020-08-20 8 1 28.5 8 1 33.7 8
2020-08-21 8 1 27.5 8 1 30.8 8
Homogeneous number.4 Minimum temperature(℃)_Niigata Quality Information.5 homogeneous number.5
2020-01-01 1 1.0 8 1
2020-01-02 1 2.0 8 1
2020-01-03 1 1.7 8 1
2020-01-04 1 2.4 8 1
2020-01-05 1 1.7 8 1
... ... ... ... ...
2020-08-17 1 26.3 8 1
2020-08-18 1 23.3 8 1
2020-08-19 1 22.0 8 1
2020-08-20 1 22.8 8 1
2020-08-21 1 23.8 8 1
[234 rows x 18 columns]
weather_parse3.jpynb
#Cut out columns
new_columns = ['Highest temperature(℃)_Tokyo', 'Highest temperature(℃)_Niigata'] #Column name
new_data=df[['Highest temperature(℃)_Tokyo','Highest temperature(℃)_Niigata']]
#recreate dataframe
new_df=pd.DataFrame(new_data,columns=new_columns)
print(new_df)
Output result
Highest temperature(℃)_東京 Highest temperature(℃)_Niigata
2020-01-01 10.2 5.6
2020-01-02 11.3 5.7
2020-01-03 12.0 6.9
2020-01-04 12.2 8.3
2020-01-05 10.2 7.2
... ... ...
2020-08-17 36.5 30.4
2020-08-18 34.3 29.0
2020-08-19 34.2 30.9
2020-08-20 34.8 33.7
2020-08-21 36.0 30.8
[234 rows x 2 columns]
weather_parse3.jpynb
new_df2=new_df['2020-08-01':'2020-08-21']
print(new_df2)
Output result
Highest temperature(℃)_Tokyo highest temperature(℃)_Niigata
2020-08-01 31.8 28.8
2020-08-02 31.5 30.3
2020-08-03 32.3 32.2
2020-08-04 33.1 32.7
2020-08-05 34.2 33.4
2020-08-06 33.1 33.6
2020-08-07 35.4 28.3
2020-08-08 32.3 24.8
2020-08-09 34.7 28.4
2020-08-10 35.2 35.6
2020-08-11 37.3 35.7
2020-08-12 35.8 31.4
2020-08-13 36.1 31.4
2020-08-14 34.2 32.0
2020-08-15 36.1 30.1
2020-08-16 35.4 31.4
2020-08-17 36.5 30.4
2020-08-18 34.3 29.0
2020-08-19 34.2 30.9
2020-08-20 34.8 33.7
2020-08-21 36.0 30.8
weather_parse3.jpynb
colorlist=["red","blue"]
#Bar graph display in specified color
new_df2.plot.bar(color=colorlist)
#Specify the display position of the legend at the bottom right
plt.legend(loc="lower right")
plt.show()
Output result
Conclusion
The maximum temperature is often higher in Tokyo.
Recommended Posts