How to plot data frame columns with exactly two unique values?

Question

I am trying to loop through a Pandas data frame and produce a bar chart only for columns that contain exactly two unique values. I envision the final bar chart to contain the two unique values on the X axis, and the Y axis to show the number of rows.

I've been able to produce a Series off my data frame (df_clean) which shows me the number of unique values per column:

    col_values = df_clean.apply(lambda x: len(x.unique()))

But I am completely lost how to:

• loop through my df_clean to only plot the columns with two unique values
• how to produce multiple graphs in one figure (I think matplotlib subplot would help?)

In the same code, I have been able to successfully loop through my df_clean and successfully plot all the int and float type columns. I am struggling with how to modify this working code for the above issue.

    i = 1
    c_num_cols = len(df_clean.select_dtypes(["int64","float64"]).columns)
    for column in df_clean.select_dtypes(["int64","float64"]).columns:
        plt.subplot(c_num_cols,(c_num_cols % 2) + 1,i)
        plt.subplots_adjust(hspace=0.5)
        df_clean[column].plot(kind = 'hist', figsize = [15,c_num_cols * 4], title = column)
        i += 1

Answer 1

Try using Series.nunique and Series.value_counts :

binary_cols = df.nunique()[lambda x: x == 2].index

for i, col in enumerate(binary_cols):
    plt.subplot(len(binary_cols), (len(binary_cols) % 2) + 1, i+1)
    plt.subplots_adjust(hspace=0.5)
    df[col].value_counts().plot(kind='bar')

---

Example

# Setup 
df = pd.DataFrame({'col1': list('aaaaaaabbbbbbbb'),
                   'col2': list('aaabbbcccdddeee'),
                   'col3': [1] * 9 + [3] * 6})


binary_cols = df.nunique()[lambda x: x == 2].index

for i, col in enumerate(binary_cols):
    plt.subplot(len(binary_cols), (len(binary_cols) % 2) + 1, i+1)
    plt.subplots_adjust(hspace=0.5)
    df[col].value_counts().plot(kind='bar')