5.4.16
Ek plot basically numbers ko pixels mein convert karna hai. Har Matplotlib chart ek hi sawaal ka jawaab deta hai:
"Main apna data 2D canvas par ink mein kaise banana chahta hoon?"
Ordered axis par trend compare karna → line Individual (x, y) points compare karna bina kisi implied connection ke → scatter Categories compare karna → bar Ek variable ki distribution / shape dekhna → histogram
Scalar field z = f ( x , y ) ko level curves ke roop mein dikhana → contour 2D array / image / matrix ko colored cells ke roop mein dikhana → imshow
80/20 rule: 90% real plots line + scatter + histogram hote hain. Inhe master karo, baaki sab variations hain.
KYA cheez inhe alag karti hai: ye aapke data ke baare mein kya assume karte hain.
Plot
Data shape
Implied meaning
Key question
ploty vs ordered x
points connected hain (ek function/series)
"trend?"
scatter(x, y) pairs
points independent hain
"correlation/cluster?"
barcategory → value
discrete groups
"sabse bada kaunsa hai?"
histsamples ka ek array
value ranges ki frequency
"distribution kaisi hai?"
contourgrid Z[i,j]=f(x,y)
ek height map (level curves)
"surface ki shape?"
imshow2D array of values
colored pixel grid
"matrix/image khud?"
Line plot ek dot-to-dot drawing hai. Aap ordered points dete ho aur Matplotlib consecutive points ke beech seedhe segments banata hai. "Curve" ki smoothness ek illusion hai jo bahut saare close points use karne se banti hai.
import numpy as np import matplotlib.pyplot as plt
x = np.linspace( 0 , 2 * np.pi, 200 ) # 200 pts -> smooth y = np.sin(x) fig, ax = plt.subplots() ax.plot(x, y, color = "tab:blue" , lw = 2 , label = "sin x" ) ax.set_xlabel( "x (rad)" ); ax.set_ylabel( "sin x" ) ax.legend(); ax.grid( True ) plt.show()
Worked example Ye step kyun?
np.linspace(0, 2π, 200) — Kyun? evenly spaced x taaki segments uniform hon; smoothness ke liye 200.lw=2 — Kyun? line width, purely visual hai.set_xlabel/ylabel — Kyun? unlabeled axis unreadable hoti hai; hamesha units ke saath label karo.
Scatter kehta hai "ye points CONNECTED nahi hain." Ise tab use karo jab x-order meaningless ho ya jab aap har point ka apna size/color chahte ho (ek 3rd/4th dimension s= aur c= ke zariye).
rng = np.random.default_rng( 0 ) x = rng.normal( size = 100 ) y = 2 * x + rng.normal( size = 100 ) # correlated ax.scatter(x, y, s = 30 , c = x, cmap = "viridis" , alpha = 0.7 )
Worked example Ye step kyun?
c=x — Kyun? color ek value encode karta hai → aap ek teesra variable 2D plot par le aate ho (dual coding).alpha=0.7 — Kyun? transparency overlapping dense regions dikhata hai.
labels = [ "A" , "B" , "C" ]; vals = [ 5 , 9 , 3 ] ax.bar(labels, vals, width = 0.6 , color = "tab:orange" ) Categorical x KYU: Matplotlib category k k
Histogram aapke data ka bar chart NAHI hai — ye ek bar chart hai ki kitne samples har value-range (bin) mein aate hain . Ye individual values ko discard kar deta hai aur frequency rakhta hai.
data = rng.normal( loc = 0 , scale = 1 , size = 5000 ) ax.hist(data, bins = 40 , density = True , color = "tab:green" , alpha = 0.8 )
Worked example Ye step kyun?
bins=40 — Kyun? bahut kam bins shape hide karte hain, bahut zyada spiky noise banate hain. Rule of thumb B ≈ n density=True — Kyun? same scale par theoretical PDF overlay karne deta hai.
Ek pahaad z = f ( x , y ) z = c 1 , c 2 , … contour line hai (jaise map par elevation lines). Lines paas paas = steep.
xs = np.linspace( - 3 , 3 , 100 ); ys = np.linspace( - 3 , 3 , 100 ) X, Y = np.meshgrid(xs, ys) Z = np.exp( - (X ** 2 + Y ** 2 )) # a Gaussian bump cs = ax.contour(X, Y, Z, levels = 8 , cmap = "plasma" ) ax.clabel(cs, inline = True ) # label each level # ax.contourf(...) lines ki jagah levels ke beech color fill karta hai
imshow array khud ko colored cells ki grid ke roop mein dikhata hai: value → color. Isi tarah aap images, matrices, heatmaps display karte ho. Positions ka koi interpolation nahi — har array cell ek pixel block hai.
Common mistake imshow ke do famous traps
Origin default mein top-left hoti hai (jaise ek image), isliye row 0 UPAR hoti hai. Math matrices ke liye origin="lower" use karo. Default galat kyun lagta hai: hum sochte hain y upward badhta hai, lekin image files row 0 top par store karti hain.imshow rows = y, columns = x plot karta hai, jo intuition ke ulta lagta hai. Indexing hai Z[row, col] = Z[y, x].
ax.imshow(Z, origin = "lower" , extent = [ - 3 , 3 , - 3 , 3 ], cmap = "inferno" , aspect = "auto" ) fig.colorbar(ax.images[ 0 ], ax = ax, label = "f(x,y)" )
Worked example Ye step kyun?
extent=[-3,3,-3,3] — Kyun? array indices ko real x,y coordinates mein map karta hai pixel indices ki jagah.colorbar — Kyun? color bina value↔color ki legend ke meaningless hai.
Common mistake Common errors ko steel-man karna
"plt.plot aur ax.plot same hain." Sahi lagta hai kyunki plt.plot scripts mein kaam karta hai. Fix: plt ek hidden "current axes" use karta hai; multiple subplots ke saath aap control khote ho. Hamesha explicit ax. (object-oriented) API use karo."Histogram = bar chart." Sahi lagta hai (dono mein bars hain). Fix: bar = aapke diye har category ka ek bar; hist = bars bins hain jo ye aapke samples count karke compute karta hai ."contour mera 1D x aur y leta hai." Fix: pehle meshgrid karna padega; Z 2D shaped (len(y), len(x)) hona chahiye."Meri image ulti kyun hai?" Fix: imshow default origin="upper" row 0 top par rakhta hai; math conventions ke liye origin="lower" use karo.plt.show() bhoolna
Recall Feynman: ek 12-saal ke bacche ko samjhaoGraph paper ke baare mein socho. Line : main kuch dots plot karta hoon aur unhe join-the-dots puzzle ki tarah connect karta hoon — achha hai "din bhar temperature kaise badla" ke liye. Scatter : main sirf dots bikherta hoon, koi lines nahi — achha hai "kya lamba baccha zyada bhari hota hai?" ke liye. Bar : har cheez ke liye lambe blocks — "kisne sabse zyada goals kiye?" Histogram : main sabki height ko boxes mein sort karta hoon (140-150cm, 150-160cm...) aur count karta hoon har box mein kitne hain — group ki shape dikhata hai. Contour : ek treasure map jaisa jo rings se hills dikhata hai. imshow : main graph paper ke har chhote square ko uske number ke hisaab se color karta hoon, jaise paint-by-numbers — isi tarah photo ya heatmap kaam karta hai.
"Lovely Scientists Bring Huge Color Images" → L ine, S catter, B ar, H istogram, C ontour, I mshow.
Aur setup ke liye: "Fig, Ax, Plot, Label, Show" = har plot ka 5-step ritual.
Line ki jagah scatter kab use karte hain? Jab points independent hon (koi meaningful order/connection nahi), ya extra variables ko point color/size ke zariye encode karna ho.
Line plot smooth lagta hai — actually usse smooth kya banata hai? Bahut saare closely-spaced sample points; ye consecutive points ke beech straight segments hi hote hain.
Bar chart aur histogram mein kya fark hai? Bar = aapke diye har category ka ek bar; histogram = bars auto-computed bins hain jo count karte hain ki kitne samples har value-range mein aate hain.
B equal bins ke liye histogram bin width formula kya hai? h = ( max − min ) / B
hist mein density=True kya karta hai aur kyun?Heights ko scale karta hai taaki total area = 1 ho (PDF estimate); har height c j / ( nh )
Level c par contour line mathematically kya hai? Points ka set {( x , y ) : f ( x , y ) = c }
contour se pehle np.meshgrid kyun call karna padta hai?Contour ko Z ki zaroorat hai 2D grid ke roop mein har (x,y) pair ke upar; meshgrid X, Y coordinate tables banata hai.
imshow kya dikhata hai aur rows/cols kaise mapped hote hain?Ek 2D array ko colored cells ke roop mein; rows = y, columns = x; value → color.
imshow image ulti kyun dikh sakti hai aur fix kya hai?Default origin="upper" row 0 ko top par rakhta hai; math orientation ke liye origin="lower" use karo.
ax.plot(...) ko plt.plot(...) par prefer kyun karte hain?Object-oriented API explicit control deta hai, multiple subplots ke saath zaroori hai.
imshow mein extent kya karta hai?Array indices ko real coordinate ranges [xmin,xmax,ymin,ymax] mein map karta hai.
Koi bhi plot banane ka 5-step ritual kya hai? fig,ax = subplots; ax.
(...); labels set karo; legend/colorbar; plt.show(). every method is ax.something
draws segments between points
