1. lets-plot
  2. docs
  3. dev
  4. notebooks

Removed Data Points Messages

Small examples for checking computation messages from stats.

In [1]:
import numpy as np
import pandas as pd

from lets_plot import *

LetsPlot.setup_html()
In [2]:
df_x = pd.DataFrame({
    "x": [1, 2, np.nan, 4, 5, np.nan]
})

df_xy = pd.DataFrame({
    "x": [1, 2, np.nan, 4, 5, np.nan],
    "y": [2, np.nan, 3, 4, 5, np.nan],
    "w": [1, 1, 1, np.nan, 1, 1],
    "g": ["A", "A", "A", "B", "B", "B"]
})

df_qq2 = pd.DataFrame({
    "x": [4, np.nan, 1, 9, 6, 2, 10, np.nan, 7, 5],
    "y": [7, 1, 9, 10, 4, np.nan, 3, np.nan, 6, 5]
})

df_smooth = pd.DataFrame({
    "x": [1, 2, 3, 4, 5, 6, 7, 8],
    "y": [1, np.nan, 3, 4, 5, np.nan, 7, 8]
})

df_ridges = pd.DataFrame({
    "x": [1, 2, np.nan, 4, 1, 2, 3, np.nan],
    "y": [0, 0, 0, 0, 1, 1, 1, 1]
})

df_sampling = pd.DataFrame({
    "x": list(range(10)),
    "y": list(range(10))
})

Count / Count2d

In [3]:
ggplot(df_x, aes("x")) + geom_bar()
Out[3]:
In [4]:
ggplot(df_xy, aes("x", "y")) + geom_pie()
Out[4]:

Bin / Bin2d / Binhex

In [5]:
ggplot(df_x, aes("x")) + geom_histogram(bins=4)
Out[5]:
In [6]:
ggplot(df_xy, aes("x", "y")) + geom_bin2d(bins=[3, 3])
Out[6]:
In [7]:
ggplot(df_xy, aes("x", "y")) + geom_hex(bins=[3, 3])
Out[7]:

Dotplot / YDotplot

In [8]:
ggplot(df_x, aes("x")) + geom_dotplot(binwidth=1.0)
Out[8]:
In [9]:
ggplot(df_xy, aes("x", "y")) + geom_ydotplot(binwidth=1.0)
Out[9]:

Summary / Summary Bin

In [10]:
ggplot(df_xy, aes("g", "y")) + stat_summary()
Out[10]:
In [11]:
ggplot(df_xy, aes("x", "y")) + stat_summary_bin(bins=3)
Out[11]:

Boxplot

geom_boxplot() exercises boxplot-related stats, including outlier handling.

In [12]:
ggplot(df_xy, aes("g", "y")) + geom_boxplot()
Out[12]:

Density Family

In [13]:
ggplot(df_x, aes("x")) + geom_density()
Out[13]:
In [14]:
ggplot(df_xy, aes("x", "y")) + geom_density2d()
Out[14]:
In [15]:
ggplot(df_xy, aes("x", "y")) + geom_density2df()
Out[15]:
In [16]:
ggplot(df_xy, aes("g", "y")) + geom_violin()
Out[16]:
In [17]:
ggplot(df_xy, aes("g", "y")) + geom_sina()
Out[17]:
In [18]:
ggplot(df_ridges, aes("x", "y")) + geom_area_ridges()
Out[18]:

ECDF

In [19]:
ggplot(df_x, aes("x")) + stat_ecdf()
Out[19]:

QQ / QQ Line

In [20]:
ggplot(df_x, aes(sample="x")) + geom_qq()
Out[20]:
In [21]:
ggplot(df_x, aes(sample="x")) + geom_qq_line()
Out[21]:

QQ2 / QQ2 Line

For qq2/qq2line, a row is counted as removed only when both x and y are non-finite. In df_qq2, only row 8 has both values missing.

In [22]:
ggplot(df_qq2, aes("x", "y")) + geom_qq2()
Out[22]:
In [23]:
ggplot(df_qq2, aes("x", "y")) + geom_qq2_line()
Out[23]:

Smooth / Smooth Summary

This example should show non-finite removal. It also uses LOESS sampling (max_n=4) to check the sampling message inside stat.

In [24]:
ggplot(df_smooth, aes("x", "y")) + \
    geom_smooth(method="loess", max_n=6, seed=42, labels=smooth_labels())
Out[24]:

Point Density

In [25]:
ggplot(df_xy, aes("x", "y")) + geom_pointdensity()
Out[25]: