Seaborn & Matplotlib Visual Lab

Interactive environment to explore, compare, and export visualizations with Seaborn and Matplotlib.

Dataset

{dataset_label}

Rows

{len(df):,}

Columns

{len(df.columns):,}

Numeric features

{len(numeric_cols_all)}

Categorical features

{len(categorical_cols_all)}

Missing ratio

{missing_ratio:.1f}%

Goal: Quick health check of the current dataset and a first look at its distributions.

Goal: Build Seaborn plots by selecting columns and options. The code snippet updates automatically.

PLOT SETUP

', unsafe_allow_html=True) # ------- Distribution ------- if family == "Distribution" and numeric_cols_all and num_col is not None: fig_seaborn, ax = plt.subplots(figsize=(10, 5)) if kind == "Histogram": sns.histplot( data=df, x=num_col, bins=bins, hue=hue_col, kde=False, ax=ax, log_scale=log_scale, ) elif kind == "KDE": sns.kdeplot( data=df, x=num_col, hue=hue_col, fill=True, ax=ax, log_scale=log_scale, ) elif kind == "Histogram + KDE": sns.histplot( data=df, x=num_col, bins=bins, hue=hue_col, kde=True, ax=ax, log_scale=log_scale, ) elif kind == "Box": sns.boxplot( data=df, x=num_col, ax=ax, ) elif kind == "Violin": sns.violinplot( data=df, x=num_col, ax=ax, ) else: # ECDF sns.ecdfplot( data=df, x=num_col, hue=hue_col, ax=ax, ) ax.yaxis.set_major_formatter( FuncFormatter(lambda y, _: f"{y:.0%}") ) ax.set_title(f"{kind} for {num_col}", fontsize=13, fontweight="bold") apply_dark(fig_seaborn, DARK) st.pyplot(fig_seaborn) hue_part = f', hue="{hue_col}"' if hue_col else "" extra_kwargs = "" if kind in ["Histogram", "Histogram + KDE"]: extra_kwargs = f", bins={bins}" if log_scale: extra_kwargs += ", log_scale=True" if kind in ["KDE", "ECDF"] and log_scale: extra_kwargs = ", log_scale=True" if kind == "Histogram + KDE": fn = "histplot" extra_kwargs = f", bins={bins}, kde=True" elif kind == "Histogram": fn = "histplot" elif kind == "KDE": fn = "kdeplot" elif kind == "Box": fn = "boxplot" elif kind == "Violin": fn = "violinplot" else: fn = "ecdfplot" code_str = f"""fig, ax = plt.subplots(figsize=(10, 5)) sns.{fn}(data=df, x="{num_col}"{hue_part}{extra_kwargs}, ax=ax) ax.set_title("{kind} for {num_col}") plt.show()""" description = "Distribution pattern: shape, spread, and tails of a single numeric column." # ------- Relationship ------- elif family == "Relationship" and len(numeric_cols_all) >= 2 and x_rel is not None: fig_seaborn, ax = plt.subplots(figsize=(10, 5)) if rel_kind == "Scatter": sns.scatterplot( data=df, x=x_rel, y=y_rel, hue=hue_rel, alpha=alpha_rel, s=70, ax=ax, ) elif rel_kind == "Line": sns.lineplot( data=df, x=x_rel, y=y_rel, hue=hue_rel, ax=ax, ) else: # Regression sns.regplot( data=df, x=x_rel, y=y_rel, ax=ax, scatter_kws={"alpha": alpha_rel, "s": 60}, line_kws={"linewidth": 2}, ) ax.set_title( f"{rel_kind}: {y_rel} vs {x_rel}", fontsize=13, fontweight="bold", ) apply_dark(fig_seaborn, DARK) st.pyplot(fig_seaborn) if rel_kind == "Scatter": hue_part = f', hue="{hue_rel}"' if hue_rel else "" code_str = f"""fig, ax = plt.subplots(figsize=(10, 5)) sns.scatterplot( data=df, x="{x_rel}", y="{y_rel}"{hue_part}, alpha=0.7, s=70, ax=ax, ) ax.set_title("Scatter: {y_rel} vs {x_rel}") plt.show()""" elif rel_kind == "Line": hue_part = f', hue="{hue_rel}"' if hue_rel else "" code_str = f"""fig, ax = plt.subplots(figsize=(10, 5)) sns.lineplot( data=df, x="{x_rel}", y="{y_rel}"{hue_part}, ax=ax, ) ax.set_title("Line: {y_rel} vs {x_rel}") plt.show()""" else: code_str = f"""fig, ax = plt.subplots(figsize=(10, 5)) sns.regplot( data=df, x="{x_rel}", y="{y_rel}", scatter_kws={{"alpha": 0.7, "s": 60}}, line_kws={{"linewidth": 2}}, ax=ax, ) ax.set_title("Regression: {y_rel} vs {x_rel}") plt.show()""" description = "Relationship pattern: how two numeric variables move together." # ------- Category ------- elif family == "Category" and categorical_cols_all and cat_var is not None: fig_seaborn, ax = plt.subplots(figsize=(10, 5)) df_tmp = df.copy() top_cats = ( df_tmp[cat_var] .value_counts() .head(order_top) .index ) df_tmp = df_tmp[df_tmp[cat_var].isin(top_cats)] if cat_kind == "Count": sns.countplot( data=df_tmp, y=cat_var, order=top_cats, ax=ax, ) for container in ax.containers: ax.bar_label(container, padding=3) elif cat_kind == "Bar (mean)": sns.barplot( data=df_tmp, y=cat_var, x=num_cat, order=top_cats, ax=ax, ci=95, ) elif cat_kind == "Box": sns.boxplot( data=df_tmp, y=cat_var, x=num_cat, order=top_cats, ax=ax, ) else: # Violin sns.violinplot( data=df_tmp, y=cat_var, x=num_cat, order=top_cats, ax=ax, ) ax.set_title( f"{cat_kind} for {cat_var}", fontsize=13, fontweight="bold", ) apply_dark(fig_seaborn, DARK) st.pyplot(fig_seaborn) if cat_kind == "Count": code_str = f"""fig, ax = plt.subplots(figsize=(10, 5)) sns.countplot( data=df, y="{cat_var}", ax=ax, ) ax.set_title("Count for {cat_var}") plt.show()""" elif cat_kind == "Bar (mean)": code_str = f"""fig, ax = plt.subplots(figsize=(10, 5)) sns.barplot( data=df, y="{cat_var}", x="{num_cat}", ci=95, ax=ax, ) ax.set_title("Mean {num_cat} by {cat_var}") plt.show()""" elif cat_kind == "Box": code_str = f"""fig, ax = plt.subplots(figsize=(10, 5)) sns.boxplot( data=df, y="{cat_var}", x="{num_cat}", ax=ax, ) ax.set_title("Box: {num_cat} by {cat_var}") plt.show()""" else: code_str = f"""fig, ax = plt.subplots(figsize=(10, 5)) sns.violinplot( data=df, y="{cat_var}", x="{num_cat}", ax=ax, ) ax.set_title("Violin: {num_cat} by {cat_var}") plt.show()""" description = "Category pattern: compare distributions or means across groups." # ------- Matrix / Heatmap ------- elif family == "Matrix / Heatmap" and selected_hm: corr = df[selected_hm].corr() fig_seaborn, ax = plt.subplots(figsize=(7, 6)) sns.heatmap( corr, annot=annot_hm, fmt=".2f", cmap="vlag", center=0 if center_zero else None, square=True, linewidths=1, cbar_kws={"shrink": 0.8}, ax=ax, ) ax.set_title("Correlation heatmap", fontsize=13, fontweight="bold") apply_dark(fig_seaborn, DARK) st.pyplot(fig_seaborn) center_value = "0" if center_zero else "None" code_str = f"""corr = df[{selected_hm}].corr() fig, ax = plt.subplots(figsize=(7, 6)) sns.heatmap( corr, annot={annot_hm}, fmt=".2f", cmap="vlag", center={center_value}, square=True, linewidths=1, cbar_kws={{"shrink": 0.8}}, ax=ax, ) ax.set_title("Correlation heatmap") plt.show()""" description = "Matrix pattern: scan many pairwise relationships at once." # ------- Multi-variable (pairplot) ------- elif family == "Multi-variable" and multi_vars: sample_size = min(sample_n, len(df)) cols_to_use = multi_vars + ([hue_multi] if hue_multi else []) df_sample = df[cols_to_use].dropna().sample(sample_size, random_state=42) with st.spinner("Building pairplot..."): g = sns.pairplot( df_sample, vars=multi_vars, hue=hue_multi, corner=True, diag_kind="kde", plot_kws={"alpha": 0.6}, diag_kws={"alpha": 0.7}, ) g.fig.suptitle("Pairplot", y=1.01, fontweight="bold") fig_seaborn = g.fig apply_dark(fig_seaborn, DARK) st.pyplot(fig_seaborn) code_str = f"""sample = df[{multi_vars + ([hue_multi] if hue_multi else [])}].dropna().sample({sample_n}, random_state=42) g = sns.pairplot( sample, vars={multi_vars}, hue={repr(hue_multi)}, corner=True, diag_kind="kde", plot_kws={{"alpha": 0.6}}, ) g.fig.suptitle("Pairplot", y=1.01) plt.show()""" description = "Multi-variable view: every pair of variables in one grid." st.markdown("

Goal: Build Matplotlib plots with fine-grained control on axes and layouts.

PLOT SETUP

', unsafe_allow_html=True) if mpl_type == "Line": if not numeric_cols_all: st.error("No numeric columns for line plot.") else: if x_line == "index": x_vals = np.arange(len(df)) x_label = "Index" else: x_vals = df[x_line].values x_label = x_line y_vals = df[y_line].values fig_mpl, ax = plt.subplots(figsize=(10, 5)) line_marker = None if marker == "None" else marker ax.plot(x_vals, y_vals, marker=line_marker, lw=2) ax.set_title(f"Line: {y_line} over {x_label}", fontsize=13, fontweight="bold") ax.set_xlabel(x_label) ax.set_ylabel(y_line) if use_grid: ax.grid(alpha=0.3) apply_dark(fig_mpl, DARK) st.pyplot(fig_mpl) code_mpl = f"""fig, ax = plt.subplots(figsize=(10, 5)) ax.plot( { 'np.arange(len(df))' if x_line == "index" else f'df["{x_line}"]' }, df["{y_line}"], marker={'None' if marker == "None" else repr(marker)}, lw=2, ) ax.set_title("Line: {y_line} over {x_label}") ax.set_xlabel("{x_label}") ax.set_ylabel("{y_line}") ax.grid(alpha=0.3) plt.show()""" elif mpl_type == "Scatter": if len(numeric_cols_all) < 2: st.error("No numeric columns for scatter plot.") else: fig_mpl, ax = plt.subplots(figsize=(10, 5)) if color_by: unique_vals = df[color_by].dropna().unique() cmap = plt.get_cmap("tab10") for idx, val in enumerate(unique_vals): mask = df[color_by] == val ax.scatter( df.loc[mask, x_sc], df.loc[mask, y_sc], alpha=alpha_sc, s=size_sc, label=str(val), color=cmap(idx % 10), ) ax.legend(title=color_by) else: ax.scatter( df[x_sc], df[y_sc], alpha=alpha_sc, s=size_sc, ) ax.set_title(f"Scatter: {y_sc} vs {x_sc}", fontsize=13, fontweight="bold") ax.set_xlabel(x_sc) ax.set_ylabel(y_sc) ax.grid(alpha=0.3) apply_dark(fig_mpl, DARK) st.pyplot(fig_mpl) code_mpl = f"""fig, ax = plt.subplots(figsize=(10, 5)) ax.scatter( df["{x_sc}"], df["{y_sc}"], alpha={alpha_sc}, s={size_sc}, ) ax.set_title("Scatter: {y_sc} vs {x_sc}") ax.set_xlabel("{x_sc}") ax.set_ylabel("{y_sc}") ax.grid(alpha=0.3) plt.show()""" elif mpl_type == "Bar": if cat_for_bar is None: st.error("Select a categorical column for the bar plot.") else: grouped = getattr(df.groupby(cat_for_bar)[num_for_bar], agg_bar)() grouped = grouped.sort_values(ascending=True) fig_mpl, ax = plt.subplots(figsize=(9, 5)) if horiz: ax.barh(grouped.index, grouped.values) ax.set_xlabel(num_for_bar) ax.set_ylabel(cat_for_bar) else: ax.bar(grouped.index, grouped.values) ax.set_ylabel(num_for_bar) ax.set_xlabel(cat_for_bar) plt.setp(ax.get_xticklabels(), rotation=45, ha="right") ax.set_title(f"{agg_bar} of {num_for_bar} by {cat_for_bar}", fontsize=13, fontweight="bold") ax.grid(axis="x" if horiz else "y", alpha=0.3) apply_dark(fig_mpl, DARK) st.pyplot(fig_mpl) code_mpl = f"""grouped = df.groupby("{cat_for_bar}")["{num_for_bar}"].{agg_bar}().sort_values() fig, ax = plt.subplots(figsize=(9, 5)) ax.barh(grouped.index, grouped.values) if {horiz} else ax.bar(grouped.index, grouped.values) ax.set_title("{agg_bar} of {num_for_bar} by {cat_for_bar}") plt.show()""" elif mpl_type == "Histogram": fig_mpl, ax = plt.subplots(figsize=(9, 5)) ax.hist( df[num_hist].dropna().values, bins=bins_hist, density=density_hist, alpha=0.85, ) ax.set_title(f"Histogram of {num_hist}", fontsize=13, fontweight="bold") ax.set_xlabel(num_hist) ax.set_ylabel("Density" if density_hist else "Count") ax.grid(alpha=0.3) apply_dark(fig_mpl, DARK) st.pyplot(fig_mpl) code_mpl = f"""fig, ax = plt.subplots(figsize=(9, 5)) ax.hist( df["{num_hist}"].dropna().values, bins={bins_hist}, density={density_hist}, alpha=0.85, ) ax.set_title("Histogram of {num_hist}") ax.set_xlabel("{num_hist}") ax.set_ylabel("{'Density' if density_hist else 'Count'}") ax.grid(alpha=0.3) plt.show()""" elif mpl_type == "Box": if not nums_box: st.warning("Select at least one numeric column.") else: fig_mpl, ax = plt.subplots(figsize=(10, 5)) ax.boxplot( [df[c].dropna().values for c in nums_box], labels=nums_box, vert=True, ) ax.set_title("Box plots", fontsize=13, fontweight="bold") ax.grid(alpha=0.3) apply_dark(fig_mpl, DARK) st.pyplot(fig_mpl) code_mpl = f"""fig, ax = plt.subplots(figsize=(10, 5)) ax.boxplot( [{', '.join([f'df["{c}"].dropna().values' for c in nums_box])}], labels={nums_box}, ) ax.set_title("Box plots") ax.grid(alpha=0.3) plt.show()""" else: # Subplots overview if not nums_over: st.warning("Select at least one numeric column.") else: k = len(nums_over) fig_mpl, axes = plt.subplots( 1, k, figsize=(4 * k, 4), squeeze=False, ) for idx, col_name in enumerate(nums_over): ax = axes[0, idx] data = df[col_name].dropna().values ax.hist(data, bins=30, alpha=0.8, density=True) if use_kde and len(data) > 10: x_vals = np.linspace(data.min(), data.max(), 200) kde = stats.gaussian_kde(data) ax.plot(x_vals, kde(x_vals), lw=2) ax.set_title(col_name) ax.grid(alpha=0.3) fig_mpl.suptitle("Numeric overview", fontsize=13, fontweight="bold") plt.tight_layout() apply_dark(fig_mpl, DARK) st.pyplot(fig_mpl) code_mpl = """cols = {cols} fig, axes = plt.subplots(1, len(cols), figsize=(4 * len(cols), 4), squeeze=False) for idx, name in enumerate(cols): ax = axes[0, idx] data = df[name].dropna().values ax.hist(data, bins=30, density=True, alpha=0.8) ax.set_title(name) ax.grid(alpha=0.3) plt.tight_layout() plt.show()""".format( cols=nums_over ) st.markdown("

Goal: See the same idea expressed once with Seaborn and once with Matplotlib.

', unsafe_allow_html=True) st.image(item["image"], use_container_width=True) st.markdown(f"**{item['name']}**") st.caption(item["description"]) st.caption( f"Saved at {item['timestamp'].strftime('%Y-%m-%d %H:%M')}" ) st.download_button( "Download PNG", data=item["image"], file_name=f"{item['name'].replace(' ', '_')}.png", mime="image/png", key=f"gal_dl_{item_idx}", use_container_width=True, ) st.markdown("