`geom_function()`¶

In [1]:

import pandas as pd
from scipy.stats import norm

from lets_plot import *

In [2]:

LetsPlot.setup_html()

In [3]:

ggplot() + geom_function(fun=norm.pdf, xlim=[-3, 3])

Out[3]:

1. Comparison With Empirical Density¶

In [4]:

df = pd.read_csv("https://raw.githubusercontent.com/JetBrains/lets-plot-docs/master/data/mpg.csv")

xs = df["cty"]

loc, scale = xs.mean(), xs.std()
xmin, xmax = xs.min(), xs.max()

fun = lambda t: norm.pdf(t, loc, scale)

In [5]:

ggplot(df, aes("cty")) + \
    geom_density() + \
    geom_function(fun=fun, xlim=[xmin, xmax], color="red")

Out[5]:

2. Use Data to Autodetect `xlim`¶

In [6]:

ggplot(df, aes("cty")) + \
    geom_density() + \
    geom_function(aes("cty"), data=df, fun=fun, color="red")

Out[6]:

3. Composition of Function and Statistic¶

In [7]:

ggplot(df, aes("cty")) + \
    geom_density() + \
    geom_function(fun=fun, xlim=[xmin, xmax], \
                  stat='smooth', method='loess', span=.3, color="red")

Out[7]:

4. Use Different Geometry¶

In [8]:

ggplot(df, aes("cty")) + \
    geom_density() + \
    geom_function(aes(fill='y'), fun=fun, xlim=[xmin, xmax], n=30, \
                  geom='histogram', alpha=.5) + \
    scale_fill_brewer(type='seq', palette='YlOrRd', trans='reverse')

Out[8]: