Tracer CSV de données de séries temporelles avec une valeur unixtime en Python (matplotlib)

↓ Supposons que vous souhaitiez tracer les données d'un tel fichier CSV.

$ cat 1.csv
"_time",count
1604847600,0
1604848200,2
1604848800,0
1604849400,2
1604850000,0
1604850600,3
1604851200,3
1604851800,0
1604852400,3
1604853000,1

1604847600 est une notation appelée UNIX TIME, ce qui signifie en fait ce qui suit.

2020-11-09 00:00:00,0
2020-11-09 00:10:00,2
2020-11-09 00:20:00,0

Considérez un programme qui convertit cela. .. ..

     1from datetime import datetime
     2
     3import matplotlib.pyplot as plt
     4import matplotlib.dates as md
     5import dateutil
     6
     7import sys
     8args = sys.argv
     9
    10x = []
    11y = []
    12
    13f = open(args[1], 'r')
    14
    15line = f.readline()
    16line = f.readline()
    17
    18while line:
    19    tmp = line.split(",")
    20    dt_object = datetime.fromtimestamp(int(tmp[0]))
    21
    22    x.append(str(dt_object))
    23    y.append(int(str(str(tmp[1]).replace('\n',''))))
    24
    25    line = f.readline()
    26f.close()
    27
    28counter = 0
    29for i in x:
    30    print(x[counter] + "," + str(y[counter]))
    31    counter = counter + 1

https://github.com/RuoAndo/qiita/blob/master/python-timeseries-plot/1.py

Faites une liste pour stocker les horodatages et les valeurs sur les lignes 10 et 11.

    10 x = []
    11 y = []

Convertissez UNIX TIME en date à la ligne 20.

    20    dt_object = datetime.fromtimestamp(int(tmp[0]))

Essayez de l'exécuter. .. ..

$ python 1.py 1.csv | more
2020-11-09 00:00:00,0
2020-11-09 00:10:00,2
2020-11-09 00:20:00,0
2020-11-09 00:30:00,2
2020-11-09 00:40:00,0
2020-11-09 00:50:00,3
2020-11-09 01:00:00,3
2020-11-09 01:10:00,0
2020-11-09 01:20:00,3
2020-11-09 01:30:00,1
2020-11-09 01:40:00,0
2020-11-09 01:50:00,1
2020-11-09 02:00:00,2
2020-11-09 02:10:00,2

(`) B

Après cela, créez un programme pour tracer cela.

Jetez un œil au code. ..

     1from datetime import datetime
     2
     3import matplotlib.pyplot as plt
     4import matplotlib.dates as md
     5import dateutil
     6
     7x = []
     8y = []
     9
    10f = open('1.csv', 'r')
    11
    12line = f.readline()
    13line = f.readline()
    14
    15while line:
    16    tmp = line.split(",")
    17    dt_object = datetime.fromtimestamp(int(tmp[0]))
    18
    19    x.append(str(dt_object))
    20    y.append(int(str(str(tmp[1]).replace('\n',''))))
    21
    22    line = f.readline()
    23f.close()
    24
    25counter = 0
    26for i in x:
    27    print(x[counter] + "," + str(y[counter]))
    28    counter = counter + 1
    29
    30dates = [dateutil.parser.parse(s) for s in x]
    31
    32plt.subplots_adjust(bottom=0.2)
    33plt.xticks( rotation= 80 )
    34ax=plt.gca()
    35xfmt = md.DateFormatter('%Y-%m-%d %H:%M')
    36ax.xaxis.set_major_formatter(xfmt)
    37plt.plot(dates,y)
    38plt.show()

https://github.com/RuoAndo/qiita/blob/master/python-timeseries-plot/5.py

Tracez la liste x (horodatage), y (valeur) de la 30e à la 38e ligne.

    30dates = [dateutil.parser.parse(s) for s in x]
    31
    32plt.subplots_adjust(bottom=0.2)
    33plt.xticks( rotation= 80 )
    34ax=plt.gca()
    35xfmt = md.DateFormatter('%Y-%m-%d %H:%M')
    36ax.xaxis.set_major_formatter(xfmt)
    37plt.plot(dates,y)
    38plt.show()

(`) B