使用时间戳的Python Panda多线图

Question

I am trying to draw graph using python panda, So far i am able to read sqlite database. I am not able to generate graph using timestamp. I want draw multi line graph using python panda. I want Months (X axis) vs Value (Y axis) graph for different line.

Below is my output of data frame(df):

Here is my code

import sqlite3
from pylab import *
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.dates as mdates
import datetime as dt


conn = sqlite3.connect('Metrics.db')
df = pd.read_sql("SELECT * FROM ABC", conn)

I tried using df.plot(), but it did not work.

Thank You in advance.

Answer 1

It seems like there are two separate objectives. One is to group your data by 'Type' and plot each line separately. To do that, you can use df.groupby() and then loop through groups:

for key, grp in df.groupby('Type'):
    plt.plot(grp['Timestamp'], grp['Value'])

The second question is how to format your timestamps to display the month. To do this, first convert your timestamps into datetimes, then use a formatter to print only the month name.

Here's some code that does that, combined with what we already had from the first part:

df['date'] = [pd.to_datetime(t, unit='s') for t in df['Timestamp']]
for key, grp in df.groupby('Type'):
    plt.plot(grp['date'], grp['Value'])
plt.gca().xaxis.set_major_formatter(mdates.DateFormatter('%B'))

You can adjust the argument to DateFormatter if you want to change your tick label format.

Answer 2

The following will convert the data frame index into Pandas timestamps:

import datetime

df.index = [pd.Timestamp(datetime.datetime.fromtimestamp(ts)) for ts in df.Timestamp]