Categorical Charts in Seaborn: Complete Guide

Categorical data analysis ka complete toolkit — Bar plots se lekar Swarm plots tak. Seaborn ke 7 powerful categorical chart types jo department-wise, gender-wise, aur har category-wise comparison professional level par dikhate hain.

📑 Is Guide Mein 7 Categorical Charts Cover Honge:

• sns.barplot() — Mean values + confidence intervals
• sns.countplot() — Category frequency counting
• sns.boxplot() — Distribution spread + outliers
• sns.violinplot() — Distribution shape visualization
• sns.swarmplot() — Individual data points (bee swarm)
• sns.stripplot() — Jittered individual points
• sns.catplot() — Figure-level unified categorical

🔍 Kya Hai: Seaborn barplot() matplotlib bar() se alag hai — yeh automatically mean calculate karta hai aur confidence interval (error bar) dikhata hai. Statistical comparison ke liye designed hai, raw values ke liye nahi.

💻 Basic Version:

fig, ax = plt.subplots(figsize=(10, 6))
sns.barplot(data=df, x="Department", y="Salary", ax=ax)
ax.set_title("Average Salary by Department")
plt.savefig("barplot_basic.png", dpi=150, bbox_inches="tight")
plt.show()

💻 Professional Styled Version:

fig, ax = plt.subplots(figsize=(12, 6))

sns.barplot(data=df, x="Department", y="Salary",
    hue="Gender",                  # split bars by gender
    estimator="mean",              # "mean","median","sum","std","min","max"
    errorbar="ci",                 # "ci"(95% CI),"sd","se","pi", or None
    ci=95,                         # confidence interval percentage
    palette=["#3498db", "#e74c3c"], # custom colors
    saturation=0.85,               # color saturation (0-1)
    width=0.7,                     # bar width
    edgecolor="#2c3e50",           # bar border
    linewidth=1.2,                 # border thickness
    capsize=0.1,                   # error bar cap width
    errwidth=1.5,                  # error bar line width
    dodge=True,                    # separate hue bars side-by-side
    order=["IT","Finance","Marketing","HR"],  # custom category order
    ax=ax
)

ax.set_title("Average Salary by Department & Gender", fontsize=16, fontweight="bold", pad=15)
ax.set_xlabel("Department", fontsize=12)
ax.set_ylabel("Average Salary (₹)", fontsize=12)
ax.legend(title="Gender", fontsize=10)
sns.despine()
plt.savefig("barplot_styled.png", dpi=150, bbox_inches="tight")
plt.show()

💻 Horizontal + Custom Estimator:

fig, ax = plt.subplots(figsize=(10, 6))

sns.barplot(data=df, x="Salary", y="Department",  # x,y swap = horizontal!
    estimator="median",             # median instead of mean
    errorbar="sd",                  # standard deviation bars
    palette="viridis",
    orient="h",                    # horizontal orientation
    ax=ax
)

# Data labels
for container in ax.containers:
    ax.bar_label(container, fmt="₹%.0f", padding=5, fontsize=10)

ax.set_title("Median Salary (± Std Dev)", fontsize=15, fontweight="bold")
sns.despine()
plt.savefig("barplot_horizontal.png", dpi=150, bbox_inches="tight")
plt.show()

📋 barplot() Parameters:

🔍 Kya Hai: countplot() har category mein kitne records hain woh count karke bars dikhata hai — value_counts() ka visual version. barplot() mein y-column dena padta hai, countplot() mein sirf x ya y dena hai — count automatic hota hai.

💻 Professional Styled Version:

fig, ax = plt.subplots(figsize=(10, 6))

sns.countplot(data=df, x="Department",
    hue="Gender",                  # split by gender
    palette=["#3498db", "#e74c3c"],
    edgecolor="#2c3e50",
    linewidth=1,
    saturation=0.85,
    order=df["Department"].value_counts().index,  # sorted by count
    ax=ax
)

# Data labels on bars
for container in ax.containers:
    ax.bar_label(container, fontsize=10, fontweight="bold", padding=3)

ax.set_title("Employee Count by Department & Gender", fontsize=16, fontweight="bold")
ax.set_ylabel("Count", fontsize=12)
ax.legend(title="Gender")
sns.despine()
plt.savefig("countplot.png", dpi=150, bbox_inches="tight")
plt.show()

💻 Percentage Countplot (Custom):

# Percentage labels instead of raw count
fig, ax = plt.subplots(figsize=(10, 6))

total = len(df)
sns.countplot(data=df, x="Rating", palette="RdYlGn",
    order=sorted(df["Rating"].unique()), edgecolor="white", ax=ax)

for p in ax.patches:
    pct = f"{p.get_height()/total*100:.1f}%"
    ax.text(p.get_x() + p.get_width()/2, p.get_height() + 5,
            pct, ha="center", fontweight="bold", fontsize=11)

ax.set_title("Rating Distribution (%)", fontsize=15, fontweight="bold")
sns.despine()
plt.savefig("countplot_pct.png", dpi=150, bbox_inches="tight")
plt.show()

🔍 Kya Hai: Seaborn boxplot() matplotlib version se zyada polished hai — automatic hue support, DataFrame integration, aur consistent styling. 5-number summary (min, Q1, median, Q3, max) + outliers ek chart mein.

💻 Professional Styled Version:

fig, ax = plt.subplots(figsize=(12, 6))

sns.boxplot(data=df, x="Department", y="Salary",
    hue="Gender",                  # split by gender
    palette=["#3498db", "#e74c3c"],
    width=0.6,                     # box width
    linewidth=1.5,                 # box border thickness
    fliersize=4,                   # outlier dot size
    flierprops={"marker": "o", "markerfacecolor": "#e74c3c", "alpha": 0.5},
    medianprops={"color": "#2c3e50", "linewidth": 2},
    whiskerprops={"linewidth": 1.5},
    capprops={"linewidth": 1.5},
    notch=True,                    # confidence interval notch
    showmeans=True,                # show mean marker
    meanprops={"marker": "D", "markerfacecolor": "gold", "markersize": 6},
    whis=1.5,                     # whisker length (IQR multiplier)
    showfliers=True,               # show outlier dots
    ax=ax
)

ax.set_title("Salary Distribution by Dept & Gender", fontsize=16, fontweight="bold")
ax.set_ylabel("Salary (₹)", fontsize=12)
ax.legend(title="Gender")
ax.grid(axis="y", linestyle="--", alpha=0.3)
sns.despine()
plt.savefig("boxplot_seaborn.png", dpi=150, bbox_inches="tight")
plt.show()

📋 boxplot() Parameters:

🔍 Kya Hai: Violin plot box plot + KDE combine karta hai — distribution ka actual shape dikhta hai. Bimodal distributions (2 peaks) violin mein clearly dikhti hain jo box plot mein invisible hoti hain.

💻 Professional Styled Version:

fig, ax = plt.subplots(figsize=(12, 7))

sns.violinplot(data=df, x="Department", y="Salary",
    hue="Gender",
    palette=["#3498db", "#e74c3c"],
    split=True,                    # split violin — left=Male, right=Female
    inner="quartile",              # "box","quartile","point","stick",None
    linewidth=1.5,                 # border thickness
    bw_adjust=0.8,                # KDE bandwidth
    cut=0,                         # clip at data limits
    density_norm="width",          # "area","count","width"
    saturation=0.85,
    ax=ax
)

ax.set_title("Salary Distribution Shape — Split Violin", fontsize=16, fontweight="bold")
ax.set_ylabel("Salary (₹)", fontsize=12)
ax.legend(title="Gender")
ax.grid(axis="y", linestyle="--", alpha=0.3)
sns.despine()
plt.savefig("violin_split.png", dpi=150, bbox_inches="tight")
plt.show()

💻 Violin + Box Overlay (Best of Both):

fig, ax = plt.subplots(figsize=(12, 6))

# Violin (background shape)
sns.violinplot(data=df, x="Department", y="Salary",
    inner=None, color="#d5e8f0", linewidth=0, alpha=0.6, ax=ax)

# Box (overlay details)
sns.boxplot(data=df, x="Department", y="Salary",
    width=0.15, palette="Set2", linewidth=1.5,
    showfliers=False, showmeans=True,
    meanprops={"marker": "D", "markerfacecolor": "red", "markersize": 5},
    ax=ax)

ax.set_title("Violin + Box Overlay (Shape + Summary)", fontsize=15, fontweight="bold")
sns.despine()
plt.savefig("violin_box_overlay.png", dpi=150, bbox_inches="tight")
plt.show()

📋 violinplot() Key Parameters:

🔍 Kya Hai: Swarmplot har individual data point ko dot ke roop mein dikhata hai — dots overlap nahi karte, bee swarm jaisa pattern banta hai. Small-medium datasets (<500 per category) ke liye perfect hai jahan har data point matter karta hai.

💻 Professional — Swarm + Box Overlay:

# Use smaller subset — swarm gets slow with 1000+ points
df_small = df.sample(200, random_state=42)

fig, ax = plt.subplots(figsize=(12, 6))

# Box (background)
sns.boxplot(data=df_small, x="Department", y="Salary",
    color="#ecf0f1", width=0.5, showfliers=False, linewidth=1.5, ax=ax)

# Swarm (individual dots overlay)
sns.swarmplot(data=df_small, x="Department", y="Salary",
    hue="Gender",
    palette=["#3498db", "#e74c3c"],
    size=5,                       # dot size
    alpha=0.7,                     # transparency
    edgecolor="#2c3e50",           # dot border
    linewidth=0.5,                 # dot border thickness
    dodge=True,                    # separate hue groups
    ax=ax
)

ax.set_title("Individual Salaries — Swarm + Box", fontsize=15, fontweight="bold")
ax.set_ylabel("Salary (₹)")
ax.legend(title="Gender")
sns.despine()
plt.savefig("swarm_box.png", dpi=150, bbox_inches="tight")
plt.show()

📋 swarmplot() Parameters:

🔍 Kya Hai: Stripplot swarmplot jaisa hai lekin dots randomly jitter (spread) hote hain — overlap allowed hai lekin jitter se density roughly dikhti hai. Large datasets ke liye swarmplot se better hai kyunki swarm bahut slow ho jaata hai 500+ points par.

💻 Professional — Strip + Violin Overlay:

fig, ax = plt.subplots(figsize=(12, 6))

# Violin (background)
sns.violinplot(data=df, x="Department", y="Salary",
    inner=None, color="#d5e8f0", alpha=0.5, linewidth=0, ax=ax)

# Strip (individual dots)
sns.stripplot(data=df, x="Department", y="Salary",
    color="#3498db",
    size=3,                       # dot size
    alpha=0.4,                     # transparency (important!)
    jitter=0.3,                   # random spread amount (0=no jitter, 0.5=max)
    edgecolor="none",              # no dot border — cleaner
    ax=ax
)

ax.set_title("Salary — Violin + Strip (All 900 Points)", fontsize=15, fontweight="bold")
ax.set_ylabel("Salary (₹)")
sns.despine()
plt.savefig("strip_violin.png", dpi=150, bbox_inches="tight")
plt.show()

📋 stripplot() vs swarmplot():

🔍 Kya Hai: catplot() Seaborn ka figure-level function hai jo sabhi categorical charts (bar, count, box, violin, swarm, strip, point) ko kind= parameter se control karta hai. col/row se automatic faceted panels bante hain — ek function se sab kuch.

💻 Faceted Box Plot — Department × Gender Grid:

g = sns.catplot(data=df, x="Rating", y="Salary",
    col="Department",              # separate panel per department
    kind="box",                    # "bar","count","box","violin","swarm","strip","point"
    palette="RdYlGn",
    col_wrap=2,                   # 2 panels per row
    height=4,                     # panel height
    aspect=1.3,                   # width = height × aspect
    showfliers=False,
    linewidth=1.2
)

g.fig.suptitle("Salary by Rating — Per Department", fontsize=16, fontweight="bold", y=1.03)
plt.savefig("catplot_box.png", dpi=150, bbox_inches="tight")
plt.show()

💻 Faceted Bar Plot — Row + Col:

g = sns.catplot(data=df, x="Department", y="Salary",
    row="Gender",                  # rows = Gender
    kind="bar",                    # bar plot mode
    palette="Set2",
    height=4,
    aspect=2.5,
    errorbar="sd",
    capsize=0.1
)

g.fig.suptitle("Salary by Department — Gender Rows", fontsize=16, fontweight="bold", y=1.02)
plt.savefig("catplot_bar.png", dpi=150, bbox_inches="tight")
plt.show()

💻 Point Plot (Mean + CI Line Chart):

g = sns.catplot(data=df, x="Rating", y="Salary",
    hue="Department",
    kind="point",                  # point = line + CI (like line chart for categories)
    height=5,
    aspect=1.8,
    markers=["o", "s", "^", "D"],  # different markers per hue
    linestyles=["-", "--", "-.", ":"],
    palette="Set2",
    capsize=0.1,
    errorbar="ci",
    dodge=0.3                     # spread hue groups slightly
)

g.fig.suptitle("Salary Trend by Rating & Department", fontsize=16, fontweight="bold", y=1.03)
plt.savefig("catplot_point.png", dpi=150, bbox_inches="tight")
plt.show()

📋 catplot() All kind= Options:

Next Post: Part 3

Next part mein hum cover karenge: Relationship & Regression Charts — scatterplot(), lineplot(), regplot(), lmplot(), relplot() aur statistical relationship visualization ke complete tools.

Happy Visualizing! 📊🚀