Seaborn is a library for making statistical graphics in Python. It is built on top of matplotlib and closely integrated with pandas data structures.
Data visualization Library -- Matplotlib
import matplotlib.pyplot as plt
One
Default visualization
plt.plot (): Drawing
plt.show (): display
Set parameters
plt.xticks (rotation): variable name skew
plt.xlable ('variable name'): X-axis name
plt.ylabel ('variable name '): Y-axis name
plt.title (): title name
Subgraph
fig= plt.figure (): Specifies the default drawing space
fig.add_ Subplot (4,1, x): x stands for relative position
Add note: plt.legend (LOC = 'best') 'best' represents the position of the comment box
Drawing type: fig, ax= plt.subplots (), ax is the axis of drawing, the actual drawing; fig is used to set parameters
Column chart:
ax.bar (bar_ positions, bar_ Heights, 0.5), 0.5 for column width
ax.barh (): bar chart
Scatter: ax.scatter (x, y)
Histogram (with bins)
ax.hist (y, range (4,5), bins = 20), range sets the range of values
ax.set_ Ylim (0.5): set Y-axis range
Box chart: ax.boxplot ()
Add text to an element: ax.text (x, y, ‘text’)
Visual library --
Seaborn
import seaborn as sns
One
Seaborn library is an encapsulation based on Matplotlib library
Set style
sns.set (): default style
sns.set_ Style: 5 Styles
with sns.axes_ Style: all statements under with use this style
layout
sns.despline (offset, left): set the offset and display of the axis
sns.set_ Context ("paper"): set the curve size in the graph
Palette: sns.color_ Palette () has six themes
Circular Sketchpad: sns.color_ palette(“hls”)
Paired: "paired"
Continuous type
sns.color_ Palette ("Blues"): from shallow to deep
sns.color_ palette(“Blues_ R "): from deep to shallow
sns.light_ Palette ("green"): from shallow to deep, reverse = true means from deep to shallow
sns.dark_ Palette ("green"): from deep to shallow
Line type: sns.cubehelix_ palette(8, start, rot)
Variable analysis
Univariate: sns.distplot (x, KDE, bins, fit) - histogram
x: Variable
KDN: estimation and density
Bins: divide the x-axis into several parts
Fit: distribution status
Bivariate
sns.jointplot (x, y, data) -- scatter plot
sns.jointplot (x, y, kind = "hex", color) - can reflect the data density
Multivariable
sns.pairplot (): reflect the relationship between two variables
sns.stripplot (x, y, data): not recommended. When the data volume is large, it is connected into a line
Optimization method:
sns.stripplot (x, y, data, jitter = true) - add a small amount of jitter
sns.swamplot (x, y, data) - tree
sns.violinplot (x, y, hue, data, split) - Violin diagram
Regression analysis: regplot, lmplot
sns.load_ Dataset - download data
sns.regplot (x, y, data, jitter)
Classification analysis
sns.barplot (x, y, hue, data): bar chart, depicting the overall trend
sns.pointplot (x, y, hue, data): point plot, depicting variation
sns.factorplot (x, y, data, kind): classification diagram of multi-layer panel
Facetgrid -- display subset
usage method
G= sns.FacetGrid (data, col)
g.map( plt.hist , variable)
Set parameters
g.set_ axis_ Labels(): name of the axis
g. Set (xticks, yticks): the value of X, y
G. fig.subplots_ Adjust (wspace, hspace): gap
Multivariable
g.PairGrid()
g.map_ diag()
g.map_ offdiag()
Heat map: sns.heatmap (data, vmin, vmax, center, annt,fmd, linewidth, cmap)
Annot: display numbers in each grid
FMD: display number format
Linewidth: distance between cells
CMAP: color map
