Python - Seaborn Cheat Sheet

by DarioPittera (aggialavura) via cheatography.com/83764/cs/19851/

TO START CATEGO​RICAL PLOTS (cont) MATRIX PLOTS

import seaborn as sns sns.sw​arm​plo​t(x​,y,df) swarm plot sns.he​atm​ap(​df.c​or​r())* heat map plot
# If working on a notebook - hue='categ var' divide per var - annot = True add actual values
%matpl​otlib inline
- palette='' set palette - cmap='' set a color palette

- split = True split by hue - lineco​lor='' set borders

DISTRI​BUTION PLOTS
** You can alo combine more plots by - linewi​dths=x set border width
sns.di​stp​lot​(df​['c​ol']) distri​bution plot
calling them one after each other. sns.cl​ust​erm​ap(​matrix) hierarc. clustering
- bin = x number of bins
sns.factorplot(x,y,df,kind)* - cmap='' set a color palette
- kde = False remove the line general categorical form of graph
- standa​rd_​scale = 1 normalise data
sns.jo​int​plo​t(x​,y,df) plot 2 variables TIP: when you call a plot function, press
Heat map plot needs a correl​ation matrix, or
- kind = '' kind of plot* "​shift + tab" to show the parameters
more generally, a matrix. You can use the
sns.pa​irp​lot(df) plot all vars combin needed.
pivot_​tab​le(​ind​ex,​col​umn​s,v​alues) function
estimator= can be, mean, std, or whatever
- hue='categ var' distin​guish per var to convert a dataframe.
function. It will display the bars or whatever
- palette='' set a color palette
you choose.
GRIDS
sns.ru​gpl​ot(​df[​'col']) idea of distri​bution General form, kind=: e.g., point, bar, violin,
sns.kd​epl​ot(​df[​'col']) kde plot etc. sns.pa​irp​lot(df) plot all vars
combin​ation
"​kin​d=" E.g.: hex, reg, kde.
ON CATEGO​RICAL PLOTS... - hue='categ var' divide per var

CATEGO​RICAL PLOTS What is a violin plot? - palette='' set palette

sns.ba​rpl​ot(​x,y,df) bar plot It has a similar role of a box and whisker g = sns.Pa​irG​rid(df) set (empty)
plots. It shows the distri​bution of quanti​‐ axis of pairplot
- estima​tor=''* bar values
tative data across several levels of one -g.map​(pl​t.s​catter) populate axis
sns.co​unt​plo​t(x,df) bars = count
(or more) catego​rical variables. The with some plot
sns.bo​xpl​ot(​x,y​,,df) box plot violin plot features a kernel density
-g.map​_di​ag(​plt.hist) set diag plots
- hue='categ var' divide per var estimation of the underlying distri​bution.
-g.map​_up​per​(pl​t.s​catter) set upper
- palette='' set palette What is a strip plot? plots
- orient='h' horiz. plot It will draw a scatte​rplot where one -g.map_lower(sns.kdeplot) set lower plots
sns.vi​oli​npl​ot(​x,y,df) violin plot* variable is catego​rical. It is also a good
g = sns.Fa​cet​Gri​d(d​f,c,r) empty axis
complement to a box or violin plot in
- hue='categ var' divide per var -g = g.map(plt .hist, "​c") populate axis
cases where you want to show all
- palette='' set palette histograms
observ​ations along with some repres​ent​‐
sns.st​rip​plo​t(x​,y,df) bars = scatter ation of the underlying distri​bution. -..g.m​ap(​sns.di​stplot, "​c") populate axis
- jitter = True add noise with distplots
What is a swarm plot
- hue='categ var' divide per var now some more complex stuff
It is similar to a stripp​lot(), but the points
- palette='' set palette are adjusted (only along the catego​rical
- split = True split by hue axis) so that they don’t overlap. This
gives a better repres​ent​ation of the distri​‐
bution of values, although it does not
scale as well to large numbers of
observ​ations.

By DarioPittera (aggialavura) Not published yet. Sponsored by CrosswordCheats.com

Last updated 15th June, 2019. Learn to solve cryptic crosswords!
Page 1 of 2. http://crosswordcheats.com

cheatography.com/aggialavura/
www.dariopittera.com
 Python - Seaborn Cheat Sheet
by DarioPittera (aggialavura) via cheatography.com/83764/cs/19851/

GRIDS (cont) STYLE and COLOR

- hue='categ var' divide per var sns.se​t_s​tyl​e('​dar​‐ apply darkgrid

-g = g.map(​plt.sc​atter, "​c", "​c").a​dd​_le​gend() kgrid') style

g = sns.Jo​int​Gri​d(x​,y,df) sns.se​t_s​tyl​e('​ticks') apply ticks style

general form of jointp​lot() sns.de​spine() remove borders
g = g.plot​(sn​s.r​egplot, sns.di​stplot) sns.de​spi​ne(​lef​‐ remove left border
join two plots t=True)
plt.fi​gur​e(f​igs​ize​=(x,x)) choose fig size
REGRESSION PLOTS
sns.se​t_c​ont​ext​('t​alk') set context
sns.lm​plo​t(x​,y,df) creat reg *
plot
sns.set_context set font size
- hue='categ var' divide per (font_scale)
var
.set_c​ont​ext​(''): e.g.: paper, poster, talk,
- palette='' set palette notebook, etc.
- markers='' * set mark
shape
- scatte​r_k​ws=​'dict' * set marker
size
sns.lm​plo​t(x​,y,df,col) create a
grid plot
sns.lmplot(x,y,df,row,col) X*X grid
sns.lmplot(x,y,df,row,col,hue) X*X*X grid
- aspect = x choose ratio
- size = x set size

marker​s='': e.g., o,v,etc.

scatte​r_k​ws='' e.g.: {'s':100}, it is a call to
matplo​tlib. It will be hard to remember how
to use these special cases, so no worries,
you will have a look online.

By DarioPittera (aggialavura) Not published yet. Sponsored by CrosswordCheats.com

Last updated 15th June, 2019. Learn to solve cryptic crosswords!
Page 2 of 2. http://crosswordcheats.com

cheatography.com/aggialavura/
www.dariopittera.com