仅绘制大熊猫数据框中的唯一行

Question

I have a pandas dataframe of 434300 rows with the following structure:

       x    y        p1  p2 
1      8.0  1.23e-6  10  12
2      7.9  4.93e-6  10  12
3      7.8  7.10e-6  10  12
...
.
...
4576   8.0  8.85e-6  5   16
4577   7.9  2.95e-6  5   16
4778   7.8  3.66e-6  5   16
...
...
...
434300 ...

with the key point being that for every block of varying x,y data there are p1 and p2 that do not vary . Note that these blocks of constant p1,p2 are of varying length so it is not simply a matter of slicing the data every n rows.

I would like to plot the values p1 vs p2 in a graph, but would only like to plot the unique points.

If i do plot p1 vs p2 using:

In [1]: fig=plt.figure()
In [2]: ax=plt.subplot(111)
In [3]: ax.plot(df['p1'],df['p2'])
In [4]: len(ax.lines[0].get_xdata())
Out[4]: 434300

I see that matplotlib is plotting each individual line of data which is to be expected.

What is the neatest way to plot only the unique points from columns p1 and p2?

Here is a csv of a small example dataset that has all of the important features of my dataset.

Answer 1

只需删除重复项并绘图：

df.drop_duplicates(how='all', columns=['p1', 'p2'])[['p1', 'p2]].plot()

Answer 2

You can slice the p1 and p2 columns from the data frame and then drop duplicates before plotting.

sub_df = df[['p1','p2']].drop_duplicates()
fig, ax = plt.subplots(1,1)
ax.plot(sub_df['p1'],sub_df['p2'])

Answer 3

import pandas as pd
import matplotlib.pyplot as plt

data = pd.read_csv('exampleData.csv')

d = data[['p1', 'p2']].drop_duplicates()

plt.plot(d['p1'], d['p2'], 'o')
plt.show()

Answer 4

After looking at this answer to a similar question in R (which is what the pandas dataframes are based on) I found the pandas function pandas.Dataframe.drop_duplicates . If we modify my example code as follows:

In [1]: fig=plt.figure()
In [2]: ax=plt.subplot(111)
In [3]: df.drop_duplicates(subset=['p1','p2'],inplace=True)
In [3]: ax.plot(df['p1'],df['p2'])
In [4]: len(ax.lines[0].get_xdata())
Out[4]: 15

We see that this restricts df to only the unique points to be plotted. An important point is that you must pass a subset to drop_duplicates so that it only uses those columns to determine duplicate rows.