Pandas dataframe scatter plot with 2-level Multiindex as axes

Question

I have a dataframe df with a 2-level Multiindex. I want a scatter plot with level 0 on the x-axis and level 1 on the y axis and scattered dots for all combinations which satisfy a condition, say have a nonzero value in a specific column 'col' .

import matplotlib.pyplot as plt
from itertools import product
import numpy as np

lengths = [3, 2]
df_index = pd.MultiIndex.from_product([list(product([-1,1], repeat=li)) for li in lengths], names=['level1', 'level2'])

df_cols = ['cols']
df = pd.DataFrame([[0.] * len(df_cols)] * len(df_index), index=df_index, columns=df_cols)
df['cols'] = np.random.randint(0, 2, size = len(df))
df

yields a dataframe of the following form

                       cols
level1       level2        
(-1, -1, -1) (-1, -1)     0
             (-1, 1)      0
             (1, -1)      0
             (1, 1)       0
(-1, -1, 1)  (-1, -1)     1
             (-1, 1)      0
             (1, -1)      1
             (1, 1)       1
(-1, 1, -1)  (-1, -1)     0
             (-1, 1)      0
             (1, -1)      0
             (1, 1)       0
(-1, 1, 1)   (-1, -1)     0
             (-1, 1)      0
             (1, -1)      1
             (1, 1)       0
(1, -1, -1)  (-1, -1)     0
             (-1, 1)      0
             (1, -1)      1
             (1, 1)       1
(1, -1, 1)   (-1, -1)     0
             (-1, 1)      1
             (1, -1)      1
             (1, 1)       0

...

Now, I want a scatter plot with the level1 index on the x-axis and the level2 index on the y-axis such that for every (x,y) with cols(x,y).= 0 there is a dot.

Answer 1

Let's first create an example dataframe with 2-level Multiindex:

import pandas as pd
import numpy as np
iterables = [[1, 2, 3, 4], [0,1, 2, 3, 4,5]]
my_multiindex=pd.MultiIndex.from_product(iterables, names=['first', 'second'])
series1 = pd.Series(np.random.randn(24), index=my_multiindex)
series2 = pd.Series(np.random.randn(24), index=my_multiindex)
df=pd.DataFrame({'col1':series1,'col2':series2})

Now, let's get the index values that satisfy a given condition:

index_values=df[df.col1<0].index.values

We then separate x and y coordinates:

xs=[a[0] for a in index_values]
ys=[a[1] for a in index_values]

We then plot:

from matplotlib import pyplot as plt
plt.scatter(xs,ys)

If you want the size of the scatter dots to reflect the actual values, you can use:

column_values=abs(df[df.col1<0].col1.values)
plt.scatter(xs,ys,s=column_values*10)

Edit to reflect the edited question :

You would just need to convert your xs and ys to strings. I am also using a large figure so that axis tick labels don't overlap:

plt.figure(figsize=(10,10))
plt.scatter([str(a) for a in xs],[str(a) for a in ys])