Building charts with value_counts

Question

I have the following dataframe:

       df_Mechanics = pd.DataFrame({'Car Plate End': [749, 749, 749,749, 749, 
                                                      749, 749, 749, 749, 49],
                     'Car Model': ['Mustang', 'Mustang', 'Mustang', 'Mustang', 'Mustang',
                                   'Mustang', 'Mustang', 'Mustang', 'Mustang', 'Mustang'],
                     'Replaced part': ['brake pad', 'wheel', 'engine', 'engine', 'engine',
                                       'wheel', 'engine','engine', 'engine', 'engine']
                    })

       print(df_Mechanics)
       Car Plate End    Car Model   Replaced part
            749           Mustang   brake pad
            749           Mustang   wheel
            749           Mustang   engine
            749           Mustang   engine
            749           Mustang   engine
            749           Mustang   wheel
            749           Mustang   engine
            749           Mustang   engine
            749           Mustang   engine
            749           Mustang   engine

I would like to plot a probability plot of parts that have been used. So, I did the following:

       prob = df_Mechanics['Replaced part'].value_counts(normalize=True)



       threshold = 0.05
       mask = prob > threshold

       tail_prob = prob.loc[~mask].sum()
       prob = prob.loc[mask]
       prob.plot(kind='bar')
       plt.xticks(rotation=25)
       plt.show()

The graph is as expected. But I would like to perfect your designer.

I found a very good example at this link: https://matplotlib.org/3.2.1/gallery/statistics/barchart_demo.html

I'm having trouble understanding.

I would like to make my graph horizontal, that the x-axis is up to 100% and that the probabilities appear in the bars. Thanks for listening.

score 1 · Accepted Answer · answered May 20 '20 at 12:35

1

You can use barh instead of bar for horizontal bar, then extract the patches to annotate with ax.text:

# replace prob.plot(...) with the following
ax = prob.plot(kind='barh')

# loop through the bars
for patch in ax.patches:
    # extract bar's information
    x = patch.get_width()
    bar_width=patch.get_height()
    y = patch.get_y()

    # annotate the bar
    ax.text(x-.05, y+bar_width/2, f'{x:.0%}',
            verticalalignment='center',
            color='white')

Output:

answered May 20 '20 at 12:35

Quang Hoang

146,074
10
56
74

Thank you. Do you know how to transform the x-axis up to 100%? – Jane Borges May 20 '20 at 12:50
See instruction to [PercentFormater](https://matplotlib.org/3.1.1/api/ticker_api.html#matplotlib.ticker.PercentFormatter) or in [this question](https://stackoverflow.com/questions/31357611/format-y-axis-as-percent). – Quang Hoang May 20 '20 at 12:51
I was able to change an x-axis scale as follows: ax = (prob*100).plot(kind='barh') However, the bar captions are gone. Do you know how to solve this problem? – Jane Borges May 20 '20 at 13:24
Replace `f'{x:.0%}'` with `f'{x}'`. – Quang Hoang May 20 '20 at 13:25
Worked perfectly. Note: replaces with red. Tks. – Jane Borges May 20 '20 at 13:30
Do you know how I can increase the size of the graph? tried to do: plt.figure (figsize = (3,4)). However, the graph did not increase. – Jane Borges May 20 '20 at 14:20
Pass same `figsize=(3,4)` into `prob.plot(...)` – Quang Hoang May 20 '20 at 14:21
I did not make it. Can you show how you would be in charge? – Jane Borges May 20 '20 at 14:24
1

`prob.plot(kind='barh', figzise=(10,6))`? – Quang Hoang May 20 '20 at 14:25
The graph increased completely. – Jane Borges May 20 '20 at 14:30
If I use [f'{x}'] the percentage symbol disappears. Do you know how to make him appear too? – Jane Borges May 21 '20 at 17:56
1

[here](https://stackoverflow.com/questions/45310254/fixed-digits-after-decimal-with-f-strings). – Quang Hoang May 21 '20 at 18:20

Building charts with value_counts

1 Answers1

Linked