Graphiques
Line plot
plt.plot(x,y)
python ↑
X = np.linspace(0, 4*np.pi, 1000)
Y = np.sin(X)
plt.plot(X, Y)
python ↑
from scipy.stats import norm
x = np.arange(-3, 3, 0.01)
plt.plot(x, norm.pdf(x))
plt.plot(x, norm.pdf(x, 1.0, 0.5))
python ↑
X = np.linspace(0, 4*np.pi, 50)
Y = np.sin(X)
plt.plot(X, Y, '-ok')
Horizontal & vertical lines
plt.axvline(x)
plt.axhline(y)
python ↑
x = np.arange(-3, 3, 0.01)
y = norm.pdf(x)
plt.plot(x, y)
plt.axvline(x.mean(), color='red')
plt.axhline(y.max(), color='grey')
Text
plt.text(x, y, str)
python ↑
fig, ax = plt.subplots(2, 3, sharex='col', sharey='row')
for i in range(2):
for j in range(3):
ax[i, j].text(0.5, 0.5, str((i, j)),
ha='center', va='center', fontsize=10)
Annotate
plt.annotate(xy, xtext, arrowprops, bbox, xycoords)
# xy : emplacement pointé par la flèche
# xytext : emplacement du text
# arrowprops: personnaliser la flèche
# bbox : personnaliser le cadre autour du texte
# xycoords : permet de changer le point de référence — xy sera calculé non plus à partir du coin en bas gauche (0,0) du graphique mais relativement aux coordonnées données (xcoord, ycoord).
arrowprops.arrowstyle:
arrowprops.connectionstyle:
Exemples
python ↑
# Annotation simple
x = np.arange(0, 10)
plt.plot(x, x)
plt.annotate(
'Mon commentaire\nsuper utile',
xy=(4, 4),
xytext=(2, 6),
arrowprops=dict(arrowstyle='->')
)
python ↑
# bbox, connectionstyle
x = np.arange(0, 10)
plt.plot(x, x)
plt.annotate(
'Mon commentaire\nsuper utile',
xy=(4, 4),
xytext=(6, 0),
arrowprops=dict(arrowstyle='-|>', connectionstyle="bar,angle=180,fraction=-0.2"),
bbox=dict(boxstyle="round", fc="0.8")
)
python ↑
# edge color, fill color, relpos
x = np.arange(0, 10)
plt.plot(x, x)
plt.annotate(
'Mon commentaire\nsuper utile',
xy=(4, 4),
xytext=(6, 3),
arrowprops=dict(
arrowstyle="wedge,tail_width=1.", # arrow style
fc=(1.0, 0.7, 0.7), # fill color
ec="none", # edge color
patchA=None, # head patch (définit la forme utilisée)
relpos=(0.2, 0.5) # move position
),
bbox=dict(
boxstyle="round", # box style
fc=(1.0, 0.7, 0.7), # fill color
ec="none" # edge color
)
)
python ↑
# xycoords (désigner l'emplacement d'une autre annotation)
plt.xlim(0,10)
plt.ylim(0,10)
an1 = plt.annotate('Text 1', xy=(5, 5), bbox={"fc":"w"})
an2 = plt.annotate('Text 2', xycoords=an1, xy=(1.05, 0.5), xytext=(2,0.5), va="center", bbox={"fc":"w"}, arrowprops=dict(arrowstyle="->"))
Une autre possibilité pour annoter le graphique est d’utiliser text:
python ↑
plt.xlim(0,10)
plt.ylim(0,10)
t = plt.text(
5, 5,
"Mon texte",
ha="center", va="center",
rotation=45, size=15,
bbox=dict(
boxstyle="rarrow,pad=0.3",
fc="lightgrey",
ec="k",
lw=2
))
Pie chart
plt.pie(values)
python ↑
values = [12, 55, 4, 32, 14]
explode = [0, 0, 0.2, 0, 0]
labels = ['India', 'United States', 'Russia', 'China', 'Europe']
colors = ['r', 'g', 'b', 'c', 'm']
plt.pie(values, colors=colors, labels=labels, explode=explode)
python ↑
values = [12, 55, 4, 32, 14]
labels = ['India', 'United States', 'Russia', 'China', 'Europe']
_patches, _labels, _percents = plt.pie(
values,
labels=labels,
wedgeprops=dict(width=0.5), # donut
autopct='%.0f%%', # display percentages
pctdistance=0.75 # position of percentages
)
# custom settings for percentages
plt.setp(_percents, color='white', weight='bold')
python ↑
from matplotlib.patches import Patch
values = np.array([(10,2), (15,40), (2,2), (20,12), (10,4)])
labels = ['India', 'United States', 'Russia', 'China', 'Europe']
colors = plt.get_cmap('jet')(np.linspace(0,1,5))
colorsB = np.tile(colors, 2).reshape(-1,4)
colorsB[::2, 3] = 0.3 # alpha 0.3 for even values
colorsB[1::2, 3] = 0.5 # alpha 0.5 for odd values
# Outer pie (radius 1 -> 1-width)
plt.pie(
values.sum(axis=1),
wedgeprops=dict(width=0.3, edgecolor='w'), radius=1,
labels=labels,
colors=colors
)
# Inner pie (radius 0.7 -> 0.7-width)
plt.pie(
values.flatten(),
wedgeprops=dict(width=0.3, edgecolor='w'), radius=0.7,
colors=colorsB
)
# Add legends
plt.gca().add_artist(
plt.legend(bbox_to_anchor=(1.1,1), loc='upper left')
)
plt.gca().add_artist(
plt.legend(
handles=[
Patch(facecolor=colorsB[0], label='A'),
Patch(facecolor=colorsB[1], label='B')
],
bbox_to_anchor=(1.1,0.6),
loc='upper left'
)
)
Bar chart
# Barres verticales
plt.bar(categories, values)
# Barres horizontales
plt.barh(categories, values)
python ↑
values = [12, 55, 4, 32, 14]
labels = ['India', 'United States', 'Russia', 'China', 'Europe']
plt.bar(labels, values)
python ↑
values = [12, 55, 4, 32, 14]
labels = ['India', 'United States', 'Russia', 'China', 'Europe']
colors = ['r', 'g', 'b', 'c', 'm']
# Bar chart
rects = plt.bar(range(0,5), values, color=colors, tick_label=labels)
# Add annotation on top of bars
for i,rect in enumerate(rects):
plt.annotate(
values[i],
xy=(
rect.get_x() + (rect.get_width() / 2),
rect.get_y() + rect.get_height() + 1),
horizontalalignment='center'
)
# Add margin for extra space
plt.margins(y=0.10)
python ↑
# Tweaks to display bars side by side
w = .2
n = 3
k = (w*(1-n)/2)
# Values
x = np.arange(1, 11)
y = np.random.uniform(1,10,10)
y1 = y+np.linspace(-1,1,10)
y2 = y+np.linspace(-2,2,10)
y3 = y+np.linspace(-3,3,10)
# Plot
plt.bar(x+k+w*0, y1, label='A', width=w, ec='k')
plt.bar(x+k+w*1, y2, label='B', width=w, ec='k')
plt.bar(x+k+w*2, y3, label='C', width=w, ec='k')
plt.legend()
plt.xticks(x)
Scatter plot
plt.scatter(x, y)
python ↑
x = np.arange(0,60)
y = x**2
plt.scatter(x,y)
python ↑
# Scatter plot
X = np.random.normal(size=100)
Y = X + np.linspace(-1,1,100)
plt.scatter(X, Y)
# Regression line
denominator = X.dot(X) - X.mean() * X.sum()
a = (X.dot(Y) - Y.mean() * X.sum()) / denominator
b = (Y.mean() * X.dot(X) - X.mean() * X.dot(Y)) / denominator
X1 = np.arange(X.min(), X.max()+1, 1)
Y1 = a*X1+b
plt.plot(X1, Y1, '-r')
python ↑
x = np.random.rand(100)
y = np.random.rand(100)
size = np.pi * (10 * np.random.rand(100))**2
colors = np.random.rand(100)
plt.scatter(x, y, s=size, c=colors, alpha=0.8, cmap='Spectral')
plt.colorbar()
Histogram
plt.hist(x)
python ↑
a = np.random.randint(20, 60, size=50)
_min = 20
_max = 60
_step = 5
plt.hist(a, bins=int((_max-_min)/_step), range=(_min,_max))
plt.grid()
python ↑
x1 = np.random.normal(27000, 15000, 10000)
x2 = np.random.normal(27000, 15000, 5000)
plt.hist(x1, bins=50, label='X1')
plt.hist(x2, bins=50, label='X2')
plt.legend()
python ↑
std = 3
mu = 0
x = np.random.normal(mu,std,1000)
n, bins, patches = plt.hist(x, bins=50, density=True)
y = ((1 / (np.sqrt(2 * np.pi) * std))
* np.exp(-0.5 * (1 / std * (bins - mu))**2))
plt.plot(bins, y, '--')
Boxplot
plt.boxplot(x)
python ↑
x = np.concatenate((
np.random.rand(100) * 100 - 40,
np.random.rand(10) * 50 + 100,
np.random.rand(10) * -50 - 100
))
plt.boxplot(x)
python ↑
x2 = x * 1.7 + 50
plt.boxplot([x, x2], labels=['A', 'B'], vert=False)
Violinplot
plt.violinplot(x)
python ↑
x = np.concatenate((
np.random.rand(100) * 100 - 40,
np.random.rand(10) * 50 + 100,
np.random.rand(10) * -50 - 100
))
plt.violinplot(x)
Errorbar
plt.errorbar(x, y, yerr)
python ↑
x = np.arange(5)
y = np.random.uniform(1,6,5)
plt.errorbar(x, y, yerr=y/4,
fmt='.k',
elinewidth=1,
capsize=2)
Fill between
plt.fill_between(x, y1, y2=0)
python ↑
x = range(1,15)
y = [1,4,6,8,4,5,3,2,4,1,5,6,8,7]
plt.plot(x,y)
plt.fill_between(x, y, color="skyblue", alpha=0.4)
python ↑
# Scatter plot
X = np.random.normal(size=100)
Y = X + np.linspace(-1,1,100)
plt.scatter(X, Y)
# Regression line
denominator = X.dot(X) - X.mean() * X.sum()
a = (X.dot(Y) - Y.mean() * X.sum()) / denominator
b = (Y.mean() * X.dot(X) - X.mean() * X.dot(Y)) / denominator
X1 = np.arange(X.min(), X.max()+1, 1)
Y1 = a*X1+b
plt.plot(X1, Y1, '-r')
# Confidence bands
Y_err = X1.std() * np.sqrt(1/len(X1) +
(X1 - X1.mean())**2 / np.sum((X1 - X1.mean())**2))
plt.fill_between(X1,
Y1 - Y_err,
Y1 + Y_err,
color="skyblue", alpha=0.4)
Contour
# Contour
plt.contour(X, Y, Z?)
# Contour + filling
plt.contourf(X, Y, Z?)
python ↑
X = np.random.uniform(0,1,(8,8))
plt.contourf(X)
python ↑
url = 'http://www.esrl.noaa.gov/psd/thredds/dodsC/Datasets/ncep.reanalysis/surface/air.sig995.2018.nc'
from netCDF4 import Dataset as NetCDFFile
nc = NetCDFFile(url)
# Extract data from NetCDF file (may take a minute)
lats = nc.variables['lat'][:]
lons = nc.variables['lon'][:]
time = nc.variables['time'][:]
air = nc.variables['air'][:] # shape is time, lat, lon
# Convert time values to datetime objects
import datetime as dt
dt_time = [dt.date(1, 1, 1) + dt.timedelta(hours=t) for t in time]
# Take a random day in the year
time_idx = 237
cur_time = dt_time[time_idx]
# Draw the world temperature for this day
plt.contourf(lons, lats, air[time_idx, :, :], 11, cmap=plt.cm.Spectral_r)
plt.title("Air temperature on %s" % cur_time)
plt.colorbar()
#-------------------------------------
print(lons.shape) # (144,)
print(lats.shape) # (73,)
print(air[time_idx, :, :].shape) # (73, 144)
# Point haut gauche
print(lons[0]) # 0.0
print(lats[0]) # 90.0
print(air[time_idx, 0, 0]) # 242.6
# Point au centre
print(lons[72]) # 180.0
print(lats[36]) # 0.0
print(air[time_idx, 36, 72]) # 300.8
# Point bas droit
print(lons[143]) # 357.5
print(lats[72]) # -90.0
print(air[time_idx, 72, 143]) # 230.6
python ↑
plt.contour(lons, lats, air[time_idx, :, :], 11, cmap=plt.cm.Spectral_r)
Imshow
plt.imshow(X)
python ↑
X = np.random.uniform(0,1,(8,8))
plt.imshow(X)
python ↑
# Load mnist
from sklearn.datasets import load_digits
mnist = load_digits()
X = mnist.data
Y = mnist.target
# Print images
i = 0
nrows = 2
ncols = 5
fig, ax = plt.subplots(
nrows,
ncols,
subplot_kw={'xticks': [], 'yticks': []},
figsize=(6,3)
)
for row in range(nrows):
for col in range(ncols):
img = X[i].reshape((8,8))
label = Y[i]
ax[row,col].imshow(img, cmap=plt.get_cmap('gray_r'))
ax[row,col].set_title('target: %d' % label)
i += 1
fig.tight_layout()
Hist2D
plt.hist2d(x, y)
python ↑
mean = [0, 0]
cov = [[1, 1], [1, 2]]
x, y = np.random.multivariate_normal(mean, cov, 10000).T
plt.hist2d(x, y, bins=30, cmap='Blues')
Hexbin
plt.hexbin(x, y)
python ↑
plt.hexbin(x, y, gridsize=30, cmap='Blues')
Step
plt.step(x, y)
python ↑
x = np.array([1, 3, 4, 5, 7])
y = np.array([1, 9, 16, 25, 49])
plt.step(x, y, where='pre', color='green', alpha=0.5)
plt.step(x, y, where='post', color='red', alpha=0.5)
plt.step(x, y, where='mid', color='blue', alpha=0.5)
Quiver
plt.quiver(U,V)
python ↑
# Base of arrows
X = [0, 0]
Y = [0, 0]
# Direction of arrows
U = [1, 0]
V = [1, -1]
# Creating plot
plt.quiver(X, Y, U, V, scale=5)
python ↑
x,y = np.meshgrid(np.arange(-2, 2, .2), np.arange(-2, 2, .25))
z = x*np.exp(-x**2 - y**2)
v, u = np.gradient(z, .2, .2)
plt.quiver(u,v)
Barbs
plt.barbs(U,V)
python ↑
x = np.linspace(0, 15, 5)
X, Y = np.meshgrid(x, x)
U, V = X**2, Y**2
plt.barbs(X, Y, U, V, U)
Eventplot
plt.eventplot(x)
python ↑
spike = 100*np.random.random(100)
plt.eventplot(spike)
Xcorr
plt.xcorr(x,y)
python ↑
x, y = np.random.randn(2, 100)
plt.xcorr(x, y, maxlags=50)
3D Projections
python ↑
a = np.arange(-1, 1, 0.02)
b = np.arange(-1, 1, 0.02)
a, b = np.meshgrid(a, b)
ax = plt.axes(projection='3d')
ax.plot_surface(a, b, a**2 + b**2)
3d plots in Python with examples
Basemap
python ↑
m = Basemap(projection='moll',
llcrnrlat=-90, urcrnrlat=90,
llcrnrlon=0, urcrnrlon=360,
resolution='c', lon_0=0)
m.drawcoastlines()
m.drawmapboundary(fill_color="#DDEEFF")
m.fillcontinents(color="#FFDDCC", lake_color='#DDEEFF')