How to add a horizontal mean line and annotation to each facet

Question

I have a simple FacetGrid of 2 row and 1 column with lineplots denoting different categories for the facets – image below.

# lineplot for each Category over the last three years
g = sns.FacetGrid(df, row="Category", sharey=False, sharex=False, height=2.5, aspect = 3)
g = g.map(plt.plot, 'Date', 'Count')

How do I add a reference line and annotation showing the mean Count for each facet?

Sample Data

Read the sample dataframe with

df = pd.read_html('https://stackoverflow.com/q/59058641/7758804')[0]
df.Date = pd.to_datetime(df.Date).dt.date

Category	Date	Count
Group 1	2017-01-31	226
Group 1	2017-02-28	235
Group 1	2017-03-31	236
Group 1	2017-04-30	221
Group 1	2017-05-31	187
Group 1	2017-06-30	218
Group 1	2017-07-31	225
Group 1	2017-08-31	221
Group 1	2017-09-30	248
Group 1	2017-10-31	224
Group 1	2017-11-30	204
Group 1	2017-12-31	224
Group 1	2018-01-31	218
Group 1	2018-02-28	241
Group 1	2018-03-31	196
Group 1	2018-04-30	246
Group 1	2018-05-31	256
Group 1	2018-06-30	217
Group 1	2018-07-31	229
Group 1	2018-08-31	230
Group 1	2018-09-30	222
Group 1	2018-10-31	215
Group 1	2018-11-30	226
Group 1	2018-12-31	227
Group 1	2019-01-31	232
Group 1	2019-02-28	233
Group 1	2019-03-31	224
Group 1	2019-04-30	214
Group 1	2019-05-31	243
Group 1	2019-06-30	214
Group 1	2019-07-31	235
Group 1	2019-08-31	218
Group 1	2019-09-30	208
Group 2	2017-01-31	208
Group 2	2017-02-28	254
Group 2	2017-03-31	223
Group 2	2017-04-30	227
Group 2	2017-05-31	245
Group 2	2017-06-30	222
Group 2	2017-07-31	226
Group 2	2017-08-31	235
Group 2	2017-09-30	225
Group 2	2017-10-31	226
Group 2	2017-11-30	258
Group 2	2017-12-31	234
Group 2	2018-01-31	257
Group 2	2018-02-28	224
Group 2	2018-03-31	228
Group 2	2018-04-30	222
Group 2	2018-05-31	227
Group 2	2018-06-30	256
Group 2	2018-07-31	217
Group 2	2018-08-31	243
Group 2	2018-09-30	230
Group 2	2018-10-31	250
Group 2	2018-11-30	197
Group 2	2018-12-31	232
Group 2	2019-01-31	248
Group 2	2019-02-28	232
Group 2	2019-03-31	259
Group 2	2019-04-30	259
Group 2	2019-05-31	229
Group 2	2019-06-30	228
Group 2	2019-07-31	234
Group 2	2019-08-31	218
Group 2	2019-09-30	231

Asked By: ZKING

||

Source

Answer 1

As per the Warning in seaborn.FacetGrid, it is better to use figure-level functions like sns.relplot.

g = sns.relplot(data=df, kind='line', x='Date', y='Count', row='Category', height=2.5, aspect=3, facet_kws={'sharey': True, 'sharex': False})
g.fig.tight_layout()

# draw lines:
for m, ax in zip(df.groupby('Category').Count.mean(), g.axes.ravel()):
    ax.hlines(m, *ax.get_xlim())
    ax.annotate(f'Mean: {m:0.0f}', xy=(ax.get_xlim()[1], m))

This also works for other figure-level plots like sns.catplot.

g = sns.catplot(data=df, kind='bar', x='Date', y='Count', row='Category', height=2.5, aspect=3)
g.set_xticklabels(rotation=90)

# draw lines:
for m, ax in zip(df.groupby('Category').Count.mean(), g.axes.ravel()):
    ax.hlines(m, *ax.get_xlim())
    ax.annotate(f'Mean: {m:0.0f}', xy=(ax.get_xlim()[1], m))

You can manually draw the horizontal line on each of the axes:

zip(list1, list2) is similar to [(list1[0], list2[0]), (list1[1], list2[1]),...]. In this code it means m is the mean, ax is the axis in the facets. ravel() turns n-dimension np.array into 1D array so you can zip. ax.hlines(y_val, x_min, x_max) draw a horizontal line at y_val from x_min to x_max. Here the two x values are provided by *ax.get_xlim().

g = sns.FacetGrid(df, row="Category", sharey=False, sharex=False, height=2.5, aspect = 3)
g = g.map(plt.plot, 'Date', 'Count')

# draw lines:
for m,ax in zip(df.groupby('Category').Count.mean(), g.axes.ravel()):
    ax.hlines(m,*ax.get_xlim())

Output:

Answered By: Quang Hoang

Answer 2

As per the Warning in seaborn.FacetGrid, it is better to use figure-level functions like sns.relplot.

g = sns.relplot(data=df, kind='line', x='Date', y='Count', row='Category', height=2.5, aspect=3, facet_kws={'sharey': True, 'sharex': False})
g.fig.tight_layout()


def custom(y, **kwargs):
    ym = y.mean()
    plt.axhline(ym, color="orange", linestyle="dashed")
    plt.annotate(f"mean: {y.mean():.3f}", xy=(1,ym), 
                 xycoords=plt.gca().get_yaxis_transform(), ha="right")
    

g = g.map(custom, 'Count')

Just a single line is enough

g = sns.FacetGrid(df, row="Category", sharey=False, sharex=False, height=2.5, aspect = 3)
g = g.map(plt.plot, 'Date', 'Count')

# add this to get a horizontal line
g = g.map(lambda y, **kw: plt.axhline(y.mean(), color="k"), 'Count')

To make the line orange and dashed and add an annotation, you could do

def custom(y, **kwargs):
    ym = y.mean()
    plt.axhline(ym, color="orange", linestyle="dashed")
    plt.annotate(f"mean: {y.mean():.3f}", xy=(1,ym), 
                 xycoords=plt.gca().get_yaxis_transform(), ha="right")
    

g = sns.FacetGrid(df, row="Category", sharey=False, sharex=False, height=2.5, aspect = 3)
g = g.map(plt.plot, 'Date', 'Count')
    
g = g.map(custom, 'Count')

Answered By: ImportanceOfBeingErnest

How to add a horizontal mean line and annotation to each facet

Question:

Sample Data

Answers: