{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Seaborn - Python 统计数据可视化教程\n",
    "\n",
    "欢迎来到 Seaborn 教程！Seaborn 是一个基于 Matplotlib 的 Python 数据可视化库。它提供了一个更高级的接口，用于绘制引人入胜且信息丰富的统计图形。\n",
    "\n",
    "**Seaborn 的优势：**\n",
    "\n",
    "1.  **专注于统计可视化**：内置了许多专门用于展示数据分布、关系和比较的统计图表类型。\n",
    "2.  **美观的默认样式**：默认的图形样式和调色板通常比 Matplotlib 的默认设置更美观。\n",
    "3.  **与 Pandas DataFrame 紧密集成**：可以方便地直接使用 Pandas DataFrame 进行绘图，通过列名指定变量。\n",
    "4.  **高级接口**：用更少的代码创建复杂的统计图形，如分面网格 (facet grids)。\n",
    "5.  **对 Matplotlib 的补充**：由于 Seaborn 构建在 Matplotlib 之上，你可以结合使用两者的功能进行深度定制。\n",
    "\n",
    "**本教程将涵盖 Seaborn 的核心绘图功能：**\n",
    "\n",
    "1.  设置与准备 (导入库, 设置样式, 加载数据)\n",
    "2.  关系图 (Relational Plots): `scatterplot`, `lineplot`\n",
    "3.  分布图 (Distribution Plots): `histplot`, `kdeplot`, `ecdfplot`\n",
    "4.  分类图 (Categorical Plots): 散点、分布、统计估计\n",
    "5.  回归图 (Regression Plots): `regplot`, `lmplot`\n",
    "6.  矩阵图 (Matrix Plots): `heatmap`, `clustermap`\n",
    "7.  风格与颜色主题\n",
    "8.  多图网格 (Figure-level interfaces & Faceting)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. 设置与准备\n",
    "\n",
    "导入必要的库，设置 Seaborn 的默认样式，并加载一些示例数据集。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "# 设置 Seaborn 的默认样式 (可选, 但推荐)\n",
    "sns.set_theme(style=\"whitegrid\") # 例如: \"whitegrid\", \"darkgrid\", \"ticks\", \"white\"\n",
    "\n",
    "# 加载 Seaborn 内置的示例数据集\n",
    "tips = sns.load_dataset(\"tips\") # 小费数据集\n",
    "iris = sns.load_dataset(\"iris\") # 鸢尾花数据集\n",
    "titanic = sns.load_dataset(\"titanic\") # 泰坦尼克号数据集\n",
    "flights = sns.load_dataset(\"flights\") # 航班乘客数据集\n",
    "\n",
    "print(\"Seaborn, Matplotlib, Pandas, NumPy imported.\")\n",
    "print(\"\\nTips dataset head:\")\n",
    "print(tips.head())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. 关系图 (Relational Plots)\n",
    "\n",
    "用于可视化两个数值变量之间的关系。\n",
    "*   `scatterplot()`: 散点图。\n",
    "*   `lineplot()`: 线图，常用于展示趋势，特别是时间序列。\n",
    "*   `relplot()`: Figure-level 接口，可以方便地创建分面网格。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"--- Scatter Plot (total_bill vs tip) ---\")\n",
    "plt.figure(figsize=(6, 4))\n",
    "sns.scatterplot(data=tips, x=\"total_bill\", y=\"tip\")\n",
    "plt.title(\"Total Bill vs Tip\")\n",
    "plt.show()\n",
    "\n",
    "# 使用 hue, size, style 添加更多维度\n",
    "print(\"\\n--- Scatter Plot with Hue, Size, Style ---\")\n",
    "plt.figure(figsize=(8, 5))\n",
    "sns.scatterplot(data=tips, x=\"total_bill\", y=\"tip\", \n",
    "                hue=\"time\",    # 按 'time' (午餐/晚餐) 区分颜色\n",
    "                size=\"size\",   # 按用餐人数 'size' 区分大小\n",
    "                style=\"smoker\" # 按是否吸烟 'smoker' 区分标记样式\n",
    "               )\n",
    "plt.title(\"Bill vs Tip (Colored by Time, Sized by Party Size, Styled by Smoker)\")\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- Line Plot (Example with dummy time series) ---\")\n",
    "# 创建一些示例时间序列数据\n",
    "time_points = np.arange(1, 31)\n",
    "values = np.random.randn(30).cumsum() + 10 # 累积和模拟趋势\n",
    "df_time = pd.DataFrame({'day': time_points, 'value': values})\n",
    "\n",
    "plt.figure(figsize=(8, 4))\n",
    "sns.lineplot(data=df_time, x=\"day\", y=\"value\", marker='o')\n",
    "plt.title(\"Example Line Plot\")\n",
    "plt.show()\n",
    "\n",
    "# 使用 relplot 创建分面网格\n",
    "print(\"\\n--- Relplot (Figure-level interface for scatter/line) ---\")\n",
    "sns.relplot(data=tips, x=\"total_bill\", y=\"tip\", \n",
    "            hue=\"smoker\", \n",
    "            col=\"time\",  # 按 'time' 分列创建子图\n",
    "            kind=\"scatter\") # 指定绘制散点图 ('line' 绘制线图)\n",
    "plt.suptitle(\"Bill vs Tip by Time and Smoker (using relplot)\", y=1.02) # 调整大标题位置\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. 分布图 (Distribution Plots)\n",
    "\n",
    "用于可视化单个变量的分布或两个变量的联合分布。\n",
    "*   `histplot()`: 直方图。\n",
    "*   `kdeplot()`: 核密度估计图 (平滑的分布曲线)。\n",
    "*   `ecdfplot()`: 经验累积分布函数图。\n",
    "*   `rugplot()`: 在坐标轴上绘制小竖线来表示每个数据点的位置。\n",
    "*   `displot()`: Figure-level 接口，可以组合多种分布图并创建分面网格。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"--- Histogram (histplot) of Total Bill ---\")\n",
    "plt.figure(figsize=(6, 4))\n",
    "sns.histplot(data=tips, x=\"total_bill\", bins=20, kde=True) # kde=True 同时绘制核密度曲线\n",
    "plt.title(\"Distribution of Total Bill\")\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- Kernel Density Estimate (kdeplot) of Tip Amount ---\")\n",
    "plt.figure(figsize=(6, 4))\n",
    "sns.kdeplot(data=tips, x=\"tip\", fill=True, color='green') # fill=True 填充曲线下方区域\n",
    "plt.title(\"Density Distribution of Tip Amount\")\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- Empirical Cumulative Distribution (ecdfplot) of Tip Amount ---\")\n",
    "plt.figure(figsize=(6, 4))\n",
    "sns.ecdfplot(data=tips, x=\"tip\")\n",
    "plt.title(\"ECDF of Tip Amount\")\n",
    "plt.ylabel(\"Cumulative Probability\")\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- Joint Distribution (histplot 2D) ---\")\n",
    "plt.figure(figsize=(6, 4))\n",
    "sns.histplot(data=tips, x=\"total_bill\", y=\"tip\", cbar=True) # 二维直方图，用颜色深浅表示密度\n",
    "plt.title(\"Joint Distribution of Total Bill and Tip\")\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- Displot (Figure-level interface) ---\")\n",
    "# kind 可以是 'hist', 'kde', 'ecdf'\n",
    "sns.displot(data=tips, x=\"total_bill\", col=\"time\", kind=\"kde\", fill=True)\n",
    "plt.suptitle(\"Distribution of Total Bill by Time (using displot)\", y=1.02)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. 分类图 (Categorical Plots)\n",
    "\n",
    "用于可视化一个数值变量与一个或多个分类变量之间的关系。\n",
    "\n",
    "*   **分类散点图 (Categorical Scatter Plots):**\n",
    "    *   `stripplot()`: 简单的散点图，点可能重叠。\n",
    "    *   `swarmplot()`: 散点图，点不会重叠（适合较小数据集）。\n",
    "*   **分类分布图 (Categorical Distribution Plots):**\n",
    "    *   `boxplot()`: 箱线图，显示分布的中位数、四分位数和异常值。\n",
    "    *   `violinplot()`: 小提琴图，结合了箱线图和核密度估计。\n",
    "    *   `boxenplot()`: 增强箱线图，提供关于分布形状的更多信息。\n",
    "*   **分类统计估计图 (Categorical Estimate Plots):**\n",
    "    *   `pointplot()`: 用点的位置显示中心趋势估计值，用误差线显示置信区间。\n",
    "    *   `barplot()`: 用条形的高度显示中心趋势估计值（默认均值）。\n",
    "    *   `countplot()`: 显示每个类别的计数（类似条形图，但 y 轴是计数）。\n",
    "*   `catplot()`: Figure-level 接口，用于绘制各种分类图并创建分面网格。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"--- Box Plot (day vs total_bill) ---\")\n",
    "plt.figure(figsize=(8, 5))\n",
    "sns.boxplot(data=tips, x=\"day\", y=\"total_bill\", palette=\"pastel\")\n",
    "plt.title(\"Total Bill Distribution by Day\")\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- Violin Plot (day vs tip, split by sex) ---\")\n",
    "plt.figure(figsize=(8, 5))\n",
    "sns.violinplot(data=tips, x=\"day\", y=\"tip\", hue=\"sex\", split=True)\n",
    "# split=True 将同一类别的不同 hue 分成两半显示\n",
    "plt.title(\"Tip Distribution by Day and Sex\")\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- Bar Plot (day vs total_bill - default estimator: mean) ---\")\n",
    "plt.figure(figsize=(8, 5))\n",
    "sns.barplot(data=tips, x=\"day\", y=\"total_bill\", hue=\"smoker\", errorbar=('ci', 95)) # errorbar 显示置信区间\n",
    "plt.title(\"Average Total Bill by Day and Smoker\")\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- Count Plot (counts of passengers by class) ---\")\n",
    "plt.figure(figsize=(6, 4))\n",
    "sns.countplot(data=titanic, x=\"class\", palette=\"viridis\")\n",
    "plt.title(\"Passenger Count by Class on Titanic\")\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- Catplot (Figure-level interface) ---\")\n",
    "# kind 可以是 'strip', 'swarm', 'box', 'violin', 'boxen', 'point', 'bar', 'count'\n",
    "sns.catplot(data=titanic, x=\"class\", y=\"age\", hue=\"sex\", \n",
    "            kind=\"violin\", split=True, col=\"survived\")\n",
    "plt.suptitle(\"Age Distribution by Class, Sex, and Survival Status (using catplot)\", y=1.02)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. 回归图 (Regression Plots)\n",
    "\n",
    "用于可视化两个变量之间的线性关系，并拟合一个回归模型。\n",
    "*   `regplot()`: Axes-level 接口，绘制散点图和线性回归拟合线。\n",
    "*   `lmplot()`: Figure-level 接口，功能更强，可以方便地进行分面和添加分类变量。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"--- Regression Plot (regplot: total_bill vs tip) ---\")\n",
    "plt.figure(figsize=(6, 4))\n",
    "sns.regplot(data=tips, x=\"total_bill\", y=\"tip\", \n",
    "            scatter_kws={'alpha':0.5}, # 传递给底层 scatter 的参数\n",
    "            line_kws={'color':'red'})   # 传递给底层 plot 的参数\n",
    "plt.title(\"Regression Fit for Total Bill vs Tip\")\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- LM Plot (lmplot: Faceted regression) ---\")\n",
    "# lmplot 使用 FacetGrid，功能更强大\n",
    "sns.lmplot(data=tips, x=\"total_bill\", y=\"tip\", \n",
    "           hue=\"smoker\", \n",
    "           col=\"time\", \n",
    "           markers=[\"o\", \"x\"],\n",
    "           height=5) # height 控制每个子图的大小\n",
    "plt.suptitle(\"Regression Fits by Smoker and Time (using lmplot)\", y=1.02)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6. 矩阵图 (Matrix Plots)\n",
    "\n",
    "用于可视化矩阵数据。\n",
    "*   `heatmap()`: 将矩阵数据绘制为颜色编码的网格。\n",
    "*   `clustermap()`: 绘制热力图，并使用层次聚类对行和/或列进行重新排序。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 计算相关系数矩阵 (示例)\n",
    "iris_numeric = iris.drop('species', axis=1) # 只保留数值列\n",
    "corr_matrix = iris_numeric.corr()\n",
    "print(f\"Correlation Matrix:\\n{corr_matrix}\")\n",
    "\n",
    "print(\"\\n--- Heatmap of Correlation Matrix ---\")\n",
    "plt.figure(figsize=(6, 5))\n",
    "sns.heatmap(corr_matrix, annot=True, cmap='coolwarm', fmt=\".2f\")\n",
    "# annot=True: 在单元格中显示数值\n",
    "# cmap: 颜色映射\n",
    "# fmt: 数值格式\n",
    "plt.title(\"Correlation Matrix Heatmap (Iris Dataset)\")\n",
    "plt.show()\n",
    "\n",
    "# --- Clustermap --- \n",
    "print(\"\\n--- Clustermap (Flights dataset example) ---\")\n",
    "# flights 数据集需要先透视成矩阵\n",
    "flights_pivot = flights.pivot_table(index='month', columns='year', values='passengers')\n",
    "print(f\"Flights Pivot Table (head):\\n{flights_pivot.head()}\")\n",
    "\n",
    "sns.clustermap(flights_pivot, cmap=\"viridis\", standard_scale=1)\n",
    "# standard_scale=1 对行进行标准化 (z-score)\n",
    "plt.suptitle(\"Clustermap of Monthly Flight Passengers Over Years\", y=1.02)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7. 风格与颜色主题\n",
    "\n",
    "Seaborn 提供了便捷的方式来控制图形的外观。\n",
    "*   **Themes**: `sns.set_theme()` (一次性设置风格、调色板等)。常用风格包括 `darkgrid`, `whitegrid`, `dark`, `white`, `ticks`。\n",
    "*   **Context**: `sns.set_context()` 控制图形元素的缩放比例，适用于不同场景 (如 `paper`, `notebook`, `talk`, `poster`)。\n",
    "*   **Palettes**: Seaborn 有许多内置的调色板，可以通过 `palette` 参数指定。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"--- Different Styles and Contexts ---\")\n",
    "\n",
    "sns.set_theme(style=\"ticks\", palette=\"pastel\") # 设置新的主题\n",
    "plt.figure(figsize=(6,4))\n",
    "sns.boxplot(data=tips, x=\"day\", y=\"total_bill\")\n",
    "plt.title(\"Boxplot with 'ticks' style and 'pastel' palette\")\n",
    "plt.show()\n",
    "\n",
    "sns.set_theme(style=\"darkgrid\") # 切换回暗网格\n",
    "sns.set_context(\"talk\") # 设置为适合演讲的上下文 (元素更大)\n",
    "plt.figure(figsize=(6,4))\n",
    "sns.histplot(data=tips, x=\"total_bill\", kde=True)\n",
    "plt.title(\"Histogram with 'darkgrid' style and 'talk' context\")\n",
    "plt.show()\n",
    "\n",
    "# 恢复默认设置 (可选)\n",
    "sns.set_theme(style=\"whitegrid\") \n",
    "sns.set_context(\"notebook\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 8. 多图网格 (Figure-level interfaces & Faceting)\n",
    "\n",
    "之前看到的 `relplot`, `displot`, `catplot`, `lmplot` 都是 **Figure-level** 接口。它们底层使用 Seaborn 的 `FacetGrid`, `PairGrid`, 或 `JointGrid` 对象来创建包含多个子图（分面）的图形。\n",
    "\n",
    "这些函数允许你通过 `row`, `col`, `hue` 等参数，轻松地根据数据的子集绘制多个相关的图表。\n",
    "\n",
    "*   **`FacetGrid`**: 最通用的，可以创建任意行、列的分面网格，然后将绘图函数 `map` 到每个子图上。\n",
    "*   **`PairGrid`**: 创建一个子图矩阵，用于展示数据集中多个变量两两之间的关系和每个变量自身的分布。\n",
    "    *   `pairplot()` 是 `PairGrid` 的便捷接口。\n",
    "*   **`JointGrid`**: 创建一个包含联合分布图和两个边缘分布图的图形。\n",
    "    *   `jointplot()` 是 `JointGrid` 的便捷接口。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"--- Pair Plot (pairplot for iris dataset) ---\")\n",
    "# 展示数据集中所有数值变量两两之间的关系以及每个变量的分布\n",
    "sns.pairplot(iris, hue=\"species\") # 按物种区分颜色\n",
    "plt.suptitle(\"Pair Plot of Iris Dataset (colored by species)\", y=1.02)\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- Joint Plot (jointplot for total_bill vs tip) ---\")\n",
    "# kind 可以是 'scatter', 'kde', 'hist', 'reg'\n",
    "sns.jointplot(data=tips, x=\"total_bill\", y=\"tip\", kind=\"reg\") # 中间散点+回归，边缘直方图\n",
    "plt.suptitle(\"Joint Distribution and Regression (Total Bill vs Tip)\", y=1.02)\n",
    "plt.show()\n",
    "\n",
    "print(\"\\n--- FacetGrid Example (Manual Mapping) ---\")\n",
    "# 手动使用 FacetGrid (更灵活，但也更复杂)\n",
    "g = sns.FacetGrid(tips, col=\"time\", row=\"smoker\", margin_titles=True)\n",
    "g.map(sns.histplot, \"total_bill\", bins=10, color='skyblue') # 将 histplot 应用到每个子图\n",
    "g.fig.suptitle(\"Total Bill Distribution by Time and Smoker (FacetGrid)\", y=1.03)\n",
    "g.set_axis_labels(\"Total Bill\", \"Count\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 总结\n",
    "\n",
    "Seaborn 是一个强大的统计数据可视化库，它构建在 Matplotlib 之上，提供了更高级的接口和更美观的默认样式。\n",
    "\n",
    "**关键要点：**\n",
    "*   专注于统计图形，如分布、关系、分类比较。\n",
    "*   与 Pandas DataFrame 紧密集成。\n",
    "*   提供 Axes-level 函数 (如 `scatterplot`, `histplot`, `boxplot`) 和 Figure-level 函数 (如 `relplot`, `displot`, `catplot`, `lmplot`)。\n",
    "*   Figure-level 函数方便创建分面网格 (Faceting)。\n",
    "*   可以轻松设置不同的视觉主题和上下文。\n",
    "*   仍然可以结合 Matplotlib 进行深度定制。\n",
    "\n",
    "对于探索性数据分析和生成用于报告或演示的统计图表，Seaborn 是一个非常有价值的工具。建议多尝试不同的绘图函数和参数组合，并查阅官方文档和示例库以获取更多灵感。"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 5
}