How to sort scatter plot by decreasing values of y axis using pyplot?

Question

I have a dataframe df1 containing two columns freq and RN with the data sorted according to ascending order by freq.

    In [2]: df1.head()
    Out[2]: 
       freq  RN
    147 1   181
    56  1   848
    149 1   814
    25  1   829

I want to plot a scatter plot with X axis as RN and y axis as freq where the X values are arranged in ascending order of the y values ie. I want the x axis to be arranged as 841,848,835,... as given in df1 which has been sorted according to ascending order of freq values.

Now if I write plt.scatter('RN', 'freq',data=df1) the output x axis I get is not sorted by the ascending order of freq. It is arranged in its own natural ascending order like 800,801,...,860.

Note: plt.bar('RN', 'freq',data=df1) works in the correct way as I want.

How Do I change it?

Answer 1

• If the RN column is numeric, the plot API will sort it numerically.
• This can be done if you set the RN column type to str.
  • This works best if the values in RN are unique. If they are not unique, all the freq values for a non-unique RN will be plotted together.
  • If RN is not unique, there's no way for the plot API to differential one value from another.

Non-Unique RN (a)

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

# create test data
np.random.seed(365)
data = {'freq': np.random.randint(20, size=(20,)), 'RN': np.random.randint(800, 900, size=(20,))}
df = pd.DataFrame(data)

# convert RN to a str type
df.RN = df.RN.astype(str)

# sort freq
df.sort_values('freq', ascending=True, inplace=True)

# plot
plt.scatter('RN', 'freq', data=df)

Non-Unique RN (b)

• Use pandas.DataFrame.groupby to group non-unique RNs together

# create test data
# create test data
np.random.seed(365)
data = {'freq': np.random.randint(20, size=(20,)), 'RN': np.random.randint(800, 900, size=(20,))}
df = pd.DataFrame(data)

# convert RN to a str type
df.RN = df.RN.astype(str)

# combine non-unique RN with groupby and sort by freq
dfg = df.groupby('RN', as_index=False)['freq'].sum().sort_values('freq')

# plot
plt.scatter('RN', 'freq', data=dfg)

Unique RN

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

# create test data
np.random.seed(365)
data = {'freq': np.random.randint(20, size=(20,)), 'RN': np.arange(800, 820)}
df = pd.DataFrame(data)

# convert RN to a str type
df.RN = df.RN.astype(str)

# sort `freq`
df.sort_values('freq', ascending=True, inplace=True)

# plot
plt.scatter('RN', 'freq', data=df)