How to plot multiple bar charts in python

Question

My intention is to re-create the below graph using matplotlib and seaborn:

The problem is that, The way I'm doing it I only get the graphs for Mediation. My current graph is shown below:

my code is shown below:

def save_bar_chart(title):
    filename = "response_time_summary_" + str(message_size) + "_" + str(backend_delay) + "ms.png"
    print("Creating chart: " + title + ", File name: " + filename)
    fig, ax = plt.subplots()
    fig.set_size_inches(11, 8)

    df_results = df.loc[(df['Message Size (Bytes)'] == message_size) & (df['Back-end Service Delay (ms)'] == backend_delay)]

    df_results = df_results[
        ['Message Size (Bytes)', 'Concurrent Users', '90th Percentile of Response Time (ms)', '95th Percentile of Response Time (ms)',
         '99th Percentile of Response Time (ms)']]

    df_results = df_results.set_index(['Message Size (Bytes)', 'Concurrent Users']).stack().reset_index().rename(
        columns={'level_2': 'Summary', 0: 'Response Time (ms)'})

    sns.barplot(x='Concurrent Users', y='Response Time (ms)', hue='Summary', data=df_results, ci=None)
    ax.yaxis.set_major_formatter(tkr.FuncFormatter(lambda y, p: "{:,}".format(y)))
    plt.suptitle(title)
    plt.legend(loc=2, frameon=True, title="Response Time Summary")
    plt.show()
    plt.savefig(filename)
    plt.clf()
    plt.close(fig)

The data looks like this:

This link contains the data

JoergVanAken · Answer 1 · 2019-03-06T16:17:09.630

0

You can reshape your DataFrame using melt

df_ = df[['Concurrent Users', '90th Percentile of Response Time (ms)',
     '95th Percentile of Response Time (ms)', '98th Percentile of Response Time (ms)',
     '99th Percentile of Response Time (ms)', '99.9th Percentile of Response Time (ms)']].melt('Concurrent Users')

and use the hue parameter in barplot

fig, ax = plt.subplots()
sns.barplot(x='Concurrent Users', y='value', hue=0, data=df_, ax=ax)

The new DataFrame df_ looks like this:

   Concurrent Users                                      0  value
0                50  90th Percentile of Response Time (ms)     26
1               100  90th Percentile of Response Time (ms)     51
2               200  90th Percentile of Response Time (ms)    105
3               300  90th Percentile of Response Time (ms)    158
4               500  90th Percentile of Response Time (ms)    243

And has these dtypes

Concurrent Users     int64
0                   object
value                int64

edited Mar 06 '19 at 16:17

answered Mar 06 '19 at 11:23

JoergVanAken

1,286
9
10

Unfortunately, this gives an error when executing line : sns.barplot(x='Concurrent Users', y='value', hue=0, data=df_, ax=ax). The error says : TypeError: 'int' object is not iterable – Suleka_28 Mar 06 '19 at 11:32
What's the output of `df_.dtypes`? – JoergVanAken Mar 06 '19 at 11:59
I did : df_results = df_results[ [ 'Concurrent Users', '90th Percentile of Response Time (ms)', '95th Percentile of Response Time (ms)', '99th Percentile of Response Time (ms)']].melt('Concurrent Users') and it gave data type as series but when I assign the result to "df_" it says object – Suleka_28 Mar 06 '19 at 12:15
Sorry, but you did not answer my question and `df_` can not be a `Series`. – JoergVanAken Mar 06 '19 at 15:46
Sorry it says DataFrame – Suleka_28 Mar 06 '19 at 15:59
Please see my edits. – JoergVanAken Mar 06 '19 at 16:17
Let us [continue this discussion in chat](https://chat.stackoverflow.com/rooms/189542/discussion-between-joergvanaken-and-suleka-28). – JoergVanAken Mar 06 '19 at 16:52

score 0 · Answer 2 · answered Mar 07 '19 at 21:33

0

You can try this:

df_out = df.set_index(['Concurrent Users','Scenario Name']).filter(like='Percentile').unstack()
df_out.columns = [f'{i} - {j}' for i, j in df_out.columns]
df_out.plot.bar(figsize=(15,10))

Output:

answered Mar 07 '19 at 21:33

Scott Boston

147,308
15
139
187

How to plot multiple bar charts in python

2 Answers2