Matplotlib in Python

Introduction to Matplotlib 🎨Matplotlib का परिचय 🎨Matplotlib 简介 🎨

Matplotlib is Python's base plotting library. seaborn and pandas .plot() are built on top of it. When you learn matplotlib directly, you can control every part of a chart: each line, label, tick and colour.

Matplotlib, Python की आधार plotting library है। seaborn और pandas .plot() इसी के ऊपर बने हैं। Matplotlib सीधे सीखने पर आप chart के हर हिस्से को नियंत्रित कर सकते हैं: हर line, label, tick और रंग।

Matplotlib 是 Python 的基础绘图库，seaborn 和 pandas .plot() 都构建在它之上。直接学习 matplotlib，你就能控制图表的每个部分：每条线、每个标签、每个刻度和每种颜色。

The Object Tree: Figure → Axes → ArtistsObject Tree: Figure → Axes → Artists对象树：Figure → Axes → Artists

Everything you see is an object you can change. The Figure is the canvas. Each Axes is one plot inside it. Lines, bars, text and ticks are artists that live inside an Axes.

जो कुछ दिखता है वह एक object है जिसे आप बदल सकते हैं। Figure canvas है। हर Axes उसके अंदर एक plot है। Lines, bars, text और ticks artists हैं जो Axes के अंदर रहते हैं।

你看到的一切都是可以修改的对象。Figure 是画布，每个 Axes 是其中的一个图。线、柱、文字和刻度都是位于 Axes 内部的 artist。

Setup and ImportsSetup और Imports准备与导入

The code blocks in this tutorial run one after another in the same notebook, so later blocks reuse variables from earlier ones. They use the two practice files customers.csv and daily_sales.csv — Section 10 shows how to make them.

इस tutorial के code blocks एक ही notebook में एक के बाद एक चलते हैं, इसलिए बाद वाले blocks पहले के variables दोबारा इस्तेमाल करते हैं। ये दो practice files customers.csv और daily_sales.csv इस्तेमाल करते हैं — Section 10 में इन्हें बनाने का तरीक़ा है।

本教程的代码块在同一个笔记本中依次运行，后面的代码块会复用前面的变量。它们使用两个练习文件 customers.csv 和 daily_sales.csv —— 第 10 节介绍如何生成。

# 1) Setup: imports, a style sheet and global defaults (rcParams)
import matplotlib.pyplot as plt
import matplotlib.ticker as mticker
import numpy as np, pandas as pd

plt.style.use("seaborn-v0_8-whitegrid")          # one line changes the whole look
plt.rcParams.update({
    "font.family": "DejaVu Sans",                    # has the ₹ glyph; Arial/Liberation do not
    "figure.figsize": (10, 5), "figure.dpi": 120,
    "axes.titlesize": 14, "axes.labelsize": 12,
    "xtick.labelsize": 10, "ytick.labelsize": 10,
    "legend.fontsize": 10, "lines.linewidth": 2,
    "axes.spines.top": False, "axes.spines.right": False,
})
df = pd.read_csv("customers.csv")
daily = pd.read_csv("daily_sales.csv", parse_dates=["date"])
print("matplotlib", plt.matplotlib.__version__, "| styles available:", len(plt.style.available))
print("customers:", df.shape, "| days:", len(daily))

Line Chart — Trends Over Time 📈Line Chart — समय के साथ Trends 📈折线图 —— 随时间的趋势 📈

A line chart joins points in order. Use it for time series and anything where order matters. Colour, width, style and markers can each be set on their own.

Line chart points को क्रम में जोड़ता है। Time series और जहाँ भी क्रम मायने रखे, वहाँ इसे इस्तेमाल करें। रंग, चौड़ाई, style और markers हर एक अलग से तय किए जा सकते हैं।

折线图按顺序连接各点，适用于时间序列以及任何顺序重要的数据。颜色、线宽、线型和标记都可以分别设置。

# 2) Line chart: monthly revenue vs target, shaded where revenue beats target
r25 = daily[daily["date"].dt.year == 2025]
months = r25["date"].dt.strftime("%b").unique()
revenue = r25.groupby(r25["date"].dt.month)["revenue"].median().to_numpy()   # median day: robust to the bad record
target = np.linspace(52, 63, 12)                                            # the sales plan

fig, ax = plt.subplots(figsize=(10, 5), layout="constrained")
ax.plot(months, revenue, color="#3b82f6", marker="o", markersize=5, label="Revenue (median day)")
ax.plot(months, target, color="#f59e0b", linewidth=1.5, linestyle="--", label="Target")
ax.fill_between(months, revenue, target, where=revenue > target, interpolate=True,
                alpha=0.2, color="#16a34a", label="Above target")
ax.set_title("2025: Revenue vs Target", fontweight="bold")
ax.set_xlabel("Month"); ax.set_ylabel("Revenue per day (₹ thousand)")
ax.yaxis.set_major_formatter(mticker.FuncFormatter(lambda v, _: f"₹{v:.0f}k"))
ax.legend(framealpha=0.3)
fig.savefig("line.png")
print("revenue:", revenue.round(1))
print("months above target:", list(months[revenue > target]))

Bar Charts — Comparing Categories 📊Bar Charts — Categories की तुलना 📊条形图 —— 比较类别 📊

Bar charts compare one number across categories. matplotlib can draw vertical, horizontal, grouped and stacked bars. For grouped bars, shift each group by half a bar width.

Bar charts एक संख्या की categories के बीच तुलना करते हैं। Matplotlib vertical, horizontal, grouped और stacked bars बना सकता है। Grouped bars के लिए हर group को आधी bar चौड़ाई से खिसकाएँ।

条形图用来比较不同类别的同一个数值。matplotlib 可以画竖直、水平、分组和堆叠条形图。画分组条形图时，把每组向左右移动半个柱宽。

# 3) Grouped bars (two age groups) and a sorted horizontal bar chart
df["age_group"] = np.where(df["age"] < 40, "Under 40", "40 and over")
avg = df.pivot_table(index="product_category", columns="age_group",
                     values="purchase_amount", aggfunc="mean") / 1000          # ₹ thousand
categories = avg.index.to_numpy()
total = df.groupby("product_category")["purchase_amount"].sum() / 1e6          # ₹ million

fig, axes = plt.subplots(1, 2, figsize=(12, 5), layout="constrained")
x = np.arange(len(categories)); width = 0.38
b1 = axes[0].bar(x - width / 2, avg["Under 40"], width, label="Under 40", color="#60a5fa")
b2 = axes[0].bar(x + width / 2, avg["40 and over"], width, label="40 and over", color="#f87171")
axes[0].bar_label(b1, fmt="%.1f", padding=3, fontsize=9)
axes[0].bar_label(b2, fmt="%.1f", padding=3, fontsize=9)
axes[0].set_xticks(x, categories, rotation=30, ha="right")                   # labels in one call (mpl ≥ 3.5)
axes[0].set_ylabel("Average spend (₹ thousand)"); axes[0].legend()

order = total.sort_values()
axes[1].barh(order.index, order.to_numpy(), color="#f59e0b")
axes[1].bar_label(axes[1].containers[0], fmt="₹%.1fM", padding=3)
axes[1].set_xlabel("Total revenue (₹ million)")
fig.savefig("bars.png")
print(avg.round(1)); print(order.round(2).to_string())

Scatter Plot — Four Variables at Once 🔵Scatter Plot — एक साथ चार Variables 🔵散点图 —— 同时展示四个变量 🔵

ax.scatter() can show up to four variables in one picture: x position, y position, colour (c=) and size (s=). Note that s is the marker area in points², not its width.

ax.scatter() एक तस्वीर में चार variables तक दिखा सकता है: x स्थिति, y स्थिति, रंग (c=) और आकार (s=)। ध्यान दें कि s marker का क्षेत्रफल है (points² में), चौड़ाई नहीं।

ax.scatter() 一张图最多能展示四个变量：x 位置、y 位置、颜色（c=）和大小（s=）。注意 s 是标记的面积（单位 points²），不是宽度。

# 4) Four variables in one scatter: x, y, colour and size — plus an annotation
fig, ax = plt.subplots(figsize=(9, 6), layout="constrained")
sc = ax.scatter(df["age"], df["purchase_amount"],
                c=df["rating"], cmap="viridis",          # colour = 3rd variable
                s=df["delivery_days"] * 8,              # size   = 4th variable
                alpha=0.6, edgecolors="white", linewidths=0.4)
ax.set_yscale("log")
fig.colorbar(sc, ax=ax, label="Customer rating")
top = df.loc[df["purchase_amount"].idxmax()]
ax.annotate(f"Largest order: ₹{top.purchase_amount:,.0f}",
            xy=(top.age, top.purchase_amount), xytext=(top.age - 25, top.purchase_amount * 0.9),
            arrowprops=dict(arrowstyle="->", color="#dc2626"), color="#dc2626",
            bbox=dict(boxstyle="round,pad=0.3", facecolor="#fde68a", alpha=0.4))
ax.set(xlabel="Customer age", ylabel="Purchase amount (₹, log scale)",
       title="Age vs purchase — colour = rating, size = delivery days")
fig.savefig("scatter.png")
print(f"largest order: age {top.age}, ₹{top.purchase_amount:,.0f}, {top.product_category}, "
      f"rating {top.rating:.0f}, {top.delivery_days} delivery days")

Histogram and KDE — Distributions 📉Histogram और KDE — Distributions 📉直方图与 KDE —— 分布 📉

ax.hist() returns three things: the bar heights n, the bin edges bins, and the bar objects patches. Because the bars are objects, you can recolour them one by one. With density=True the bars have area 1, so a KDE curve (from scipy.stats.gaussian_kde) fits on the same scale.

ax.hist() तीन चीज़ें लौटाता है: bar की ऊँचाइयाँ n, bin के किनारे bins, और bar objects patches। Bars objects हैं, इसलिए आप उन्हें एक-एक करके नया रंग दे सकते हैं। density=True से bars का कुल क्षेत्रफल 1 होता है, इसलिए KDE curve (scipy.stats.gaussian_kde से) उसी scale पर बैठती है।

ax.hist() 返回三样东西：柱高 n、分组边界 bins 和柱对象 patches。因为柱是对象，你可以逐个改颜色。设置 density=True 后柱的总面积为 1，因此 KDE 曲线（来自 scipy.stats.gaussian_kde）能画在同一刻度上。

# 5) Histogram + KDE, bars coloured by side of the mean; overlapping group histograms
from scipy.stats import gaussian_kde
data = df["purchase_amount"].to_numpy()
fig, axes = plt.subplots(1, 2, figsize=(12, 5), layout="constrained")

n, bins, patches = axes[0].hist(data, bins=40, range=(0, 30000), density=True,
                                color="#60a5fa", alpha=0.7, edgecolor="white", linewidth=0.4)
mean_val, med = data.mean(), np.median(data)
for patch, left in zip(patches, bins[:-1]):
    if left >= mean_val:
        patch.set_facecolor("#f59e0b")                    # bins at or right of the mean
xs = np.linspace(0, 30000, 300)
axes[0].plot(xs, gaussian_kde(data)(xs), color="#7c3aed", label="KDE")
axes[0].axvline(mean_val, color="#f59e0b", linestyle="--", label=f"Mean ₹{mean_val:,.0f}")
axes[0].axvline(med, color="#16a34a", linestyle="--", label=f"Median ₹{med:,.0f}")
axes[0].legend()

for name, colour in zip(["Low", "Medium", "High"], ["#60a5fa", "#f87171", "#34d399"]):
    vals = np.log10(df.loc[df["income_bracket"] == name, "purchase_amount"])
    axes[1].hist(vals, bins=30, alpha=0.5, color=colour, label=name, density=True)
axes[1].set_xlabel("log10(purchase amount)"); axes[1].legend(title="Income")
fig.savefig("hist.png")
print(f"mean ₹{mean_val:,.0f}  median ₹{med:,.0f}  bins at/after mean: {(bins[:-1] >= mean_val).sum()} of 40")
print("median spend:", df.groupby("income_bracket")["purchase_amount"].median().reindex(["Low", "Medium", "High"]).to_dict())

Subplots — Many Charts in One Figure 🧩Subplots — एक Figure में कई Charts 🧩子图 —— 一个 Figure 里的多张图 🧩

plt.subplots(nrows, ncols) makes a grid of Axes. It is the main layout tool for dashboards and side-by-side comparisons. Watch what it returns: the shape of axes depends on the grid size.

plt.subplots(nrows, ncols) Axes का grid बनाता है। Dashboards और साथ-साथ तुलना के लिए यह मुख्य layout tool है। ध्यान दें कि यह क्या लौटाता है: axes का shape grid के आकार पर निर्भर है।

plt.subplots(nrows, ncols) 创建一个 Axes 网格，是做仪表盘和并排比较的主要布局工具。注意它的返回值：axes 的形状取决于网格大小。

# 6) Subplots: a 2×2 dashboard, and an uneven layout with subplot_mosaic
fig, axes = plt.subplots(2, 2, figsize=(12, 8), layout="constrained")
fig.suptitle("Sales Dashboard — 2025", fontsize=16, fontweight="bold")
axes[0, 0].plot(months, revenue); axes[0, 0].set_title("Revenue trend")
axes[0, 1].barh(order.index, order.to_numpy()); axes[0, 1].set_title("Revenue by category")
axes[1, 0].hist(df["age"], bins=12); axes[1, 0].set_title("Age distribution")
axes[1, 1].scatter(df["age"], df["purchase_amount"], s=4, alpha=0.4)
axes[1, 1].set_yscale("log"); axes[1, 1].set_title("Age vs spend")
fig.savefig("dashboard.png")
print("axes array shape:", axes.shape, "| top-right is axes[0, 1]")

fig2, axd = plt.subplot_mosaic([["trend", "trend"], ["cat", "age"]], figsize=(10, 7), layout="constrained")
axd["trend"].plot(daily["date"], daily["revenue"], linewidth=0.8)     # wide panel on top
print("mosaic panels:", list(axd))

Heatmap with imshow() 🌡️imshow() से Heatmap 🌡️用 imshow() 画热力图 🌡️

ax.imshow() draws any 2-D array as a grid of coloured cells. It is the base for correlation heatmaps, pivot tables and confusion matrices. You add the numbers yourself with ax.text().

ax.imshow() किसी भी 2-D array को रंगीन cells के grid के रूप में बनाता है। Correlation heatmaps, pivot tables और confusion matrices का यही आधार है। संख्याएँ आप ax.text() से ख़ुद जोड़ते हैं।

ax.imshow() 把任意二维数组画成彩色格子网格，是相关热力图、透视表和混淆矩阵的基础。格子里的数字需要你用 ax.text() 自己加。

# 7) A correlation heatmap with ax.imshow() and a value in every cell
cols = ["age", "income_code", "purchase_amount", "delivery_days", "rating"]
corr = df[cols].corr().to_numpy(); n = len(cols)
fig, ax = plt.subplots(figsize=(7, 6), layout="constrained")
im = ax.imshow(corr, cmap="coolwarm", vmin=-1, vmax=1)
ax.grid(False)                                   # the style's grid would cut through the cells
fig.colorbar(im, ax=ax, fraction=0.046, pad=0.04)
for i in range(n):
    for j in range(n):
        ax.text(j, i, f"{corr[i, j]:.2f}", ha="center", va="center", fontweight="bold",
                color="white" if abs(corr[i, j]) > 0.6 else "black")
ax.set_xticks(range(n), cols, rotation=45, ha="right")
ax.set_yticks(range(n), cols)
ax.set_title("Correlation matrix (Pearson)")
fig.savefig("heatmap.png")
print(np.round(corr, 2))

Box Plot and Violin Plot 🎻Box Plot और Violin Plot 🎻箱线图与小提琴图 🎻

ax.boxplot() shows the median, quartiles, whiskers and outliers. ax.violinplot() adds the full shape, using a KDE drawn on both sides. Money data is very skewed, so both plots below use log10 of the amount, with ₹ tick labels added by a formatter.

ax.boxplot() median, quartiles, whiskers और outliers दिखाता है। ax.violinplot() दोनों तरफ़ बनी KDE से पूरा आकार जोड़ता है। पैसे का data बहुत skewed है, इसलिए नीचे के दोनों plots राशि का log10 इस्तेमाल करते हैं, और ₹ tick labels एक formatter से जोड़े गए हैं।

ax.boxplot() 显示中位数、四分位数、须线和异常值。ax.violinplot() 用两侧对称的 KDE 补上完整形状。金额数据非常偏，所以下面两图都用金额的 log10，并用格式化器加上 ₹ 刻度标签。

# 8) Notched box plot and violin plot of the same groups (on log10 of the amount)
cats = ["Grocery", "Books", "Clothing", "Furniture", "Electronics"]
groups = [np.log10(df.loc[df["product_category"] == c, "purchase_amount"]).to_numpy() for c in cats]
colors = ["#0d9488", "#7c3aed", "#3b82f6", "#d97706", "#ea580c"]
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 5), sharey=True, layout="constrained")

bp = ax1.boxplot(groups, tick_labels=cats, patch_artist=True, notch=True,
                 orientation="vertical", widths=0.5)   # 'vert=' is deprecated since 3.11
for patch, c in zip(bp["boxes"], colors):
    patch.set_facecolor(c); patch.set_alpha(0.6)
vp = ax2.violinplot(groups, showmedians=True)
for body, c in zip(vp["bodies"], colors):
    body.set_facecolor(c); body.set_alpha(0.5)
ax2.set_xticks(range(1, 6), cats)
ax1.yaxis.set_major_formatter(mticker.FuncFormatter(lambda v, _: f"₹{10 ** v:,.0f}"))
fig.savefig("box_violin.png")
for c, g in zip(cats, groups):
    q1, med, q3 = np.percentile(g, [25, 50, 75]); half = 1.57 * (q3 - q1) / np.sqrt(len(g))   # notch rule
    print(f"{c:<12} n={len(g):<4} median ₹{10 ** med:>6,.0f}   notch ₹{10 ** (med - half):,.0f} – ₹{10 ** (med + half):,.0f}")

Styles and Saving Figures 💾Styles और Figures Save करना 💾样式与保存图片 💾

A style sheet changes many rcParams at once. plt.style.use() changes every later chart; with plt.style.context("ggplot"): changes only the code inside the block. The picture below uses the real colours each style sets.

Style sheet एक साथ कई rcParams बदलती है। plt.style.use() बाद के हर chart को बदलता है; with plt.style.context("ggplot"): सिर्फ़ block के अंदर के code को बदलता है। नीचे की तस्वीर हर style के असली रंग इस्तेमाल करती है।

样式表一次修改许多 rcParams。plt.style.use() 影响之后所有的图；with plt.style.context("ggplot"): 只影响代码块内部。下图使用了每种样式真实设置的颜色。

# 9) Save once, in the right format
import os
fig, ax = plt.subplots(figsize=(6, 4), layout="constrained")
ax.plot(months, revenue, marker="o")
ax.set(title="Revenue 2025", ylabel="₹ thousand per day")
for name, kw in [("chart.png", dict(dpi=300)), ("chart.pdf", {}), ("chart.svg", {})]:
    fig.savefig(name, bbox_inches="tight", **kw)
    print(f"{name:<10} {os.path.getsize(name) / 1024:6.1f} KB")

Make the Practice Data 🐍Practice Data बनाएँ 🐍生成练习数据 🐍

These are the same two files used in the Data Visualisation tutorial. The patterns (older → higher income, category → spend, slow delivery → low rating, an October–November peak) are built in on purpose. Run this before the other blocks.

ये वही दो files हैं जो Data Visualisation tutorial में इस्तेमाल हुईं। Patterns (बड़ी उम्र → ज़्यादा income, category → ख़र्च, धीमी delivery → कम rating, अक्टूबर–नवंबर की चोटी) जानबूझकर डाले गए हैं। बाक़ी blocks से पहले इसे चलाएँ।

这就是“数据可视化”教程中用到的两个文件。其中的规律（年龄越大收入越高、类别决定消费、配送慢评分低、10–11 月高峰）都是有意内置的。请先运行它，再运行其他代码块。

# 0) Build two practice datasets that contain real patterns to find
import numpy as np, pandas as pd
rng = np.random.default_rng(11)

# --- customers.csv: one row per customer ---
n = 2000
age = rng.integers(18, 66, n)
p_hi = np.clip((age - 18) / 60, 0.05, 0.75)                 # older -> more likely higher income
income_code = np.minimum(3, rng.binomial(3, p_hi))
brackets = np.array(["Low", "Medium", "High", "Very High"])
cats = rng.choice(["Electronics", "Furniture", "Clothing", "Books", "Grocery"], n, p=[.15, .1, .3, .2, .25])
cat_base = pd.Series({"Electronics": 9.0, "Furniture": 8.7, "Clothing": 7.9, "Books": 7.0, "Grocery": 6.3})[cats].to_numpy()
spend = np.exp(cat_base + 0.35 * income_code + rng.normal(0, 0.45, n)).round(-1)
delivery = rng.integers(1, 11, n)
rating = np.clip(np.round(5.2 - 0.28 * delivery + rng.normal(0, 0.8, n)), 1, 5)
pd.DataFrame({"age": age, "income_bracket": brackets[income_code], "income_code": income_code,
              "product_category": cats, "purchase_amount": spend,
              "delivery_days": delivery, "rating": rating}).to_csv("customers.csv", index=False)

# --- daily_sales.csv: two years of daily revenue (₹ thousand) ---
days = pd.date_range("2024-01-01", "2025-12-31", freq="D")
t = np.arange(len(days))
trend = 40 + 20 * t / len(days)
weekly = np.where(days.dayofweek >= 5, 1.18, 1.0)
festive = 1 + 0.45 * np.exp(-((days.dayofyear - 305) / 18) ** 2) + 0.25 * np.exp(-((days.dayofyear - 358) / 6) ** 2)
revenue = np.asarray(trend * weekly * festive * rng.normal(1, 0.07, len(days)))
for d, f in [("2024-03-15", 2.2), ("2025-02-01", 5.0), ("2025-06-10", 0.2)]:   # flash sale, data error, outage
    revenue[days.get_loc(d)] *= f
pd.DataFrame({"date": days, "revenue": revenue.round(2)}).to_csv("daily_sales.csv", index=False)
print("customers.csv:", n, "rows   daily_sales.csv:", len(days), "rows")

Golden Rules of Matplotlib 🏆Matplotlib के सुनहरे नियम 🏆Matplotlib 黄金法则 🏆