plothist 1.4.0__py3-none-any.whl → 1.5.0__py3-none-any.whl

plothist/examples/advanced/1d_side_by_side_with_numbers.py

@@ -0,0 +1,81 @@
+"""
+Side-by-side categories, with numbers on top
+============================================
+
+Plot multiple 1D histograms with categories side by side, and add the number of entries on top of each bar.
+"""
+
+###
+import boost_histogram as bh
+import matplotlib.pyplot as plt
+import numpy as np
+
+from plothist import plot_hist
+
+rng = np.random.default_rng(83113111)
+
+# Integer categories
+categories = [-137, 12, 1234]
+axis = bh.axis.IntCategory(categories=categories)
+
+# Generate data for 3 histograms
+data = [
+    rng.choice(categories, 50),
+    rng.choice(categories, 30),
+    rng.choice(categories, 35),
+    rng.choice(categories, 30),
+]
+
+# Create and fill the histograms
+histos = [bh.Histogram(axis, storage=bh.storage.Weight()) for _ in range(len(data))]
+histos = [histo.fill(data[i]) for i, histo in enumerate(histos)]
+
+labels = [f"$h_{{{i}}}$" for i in range(len(histos))]
+colors = ["#348ABD", "#E24A33", "#988ED5", "#FBC15E"]
+
+# Plot the histogram
+fig, ax = plt.subplots()
+
+# Use a specificity of matplotlib: when a list of histograms is given, it will plot them side by side unless stacked=True or histtype is a "step" type.
+plot_hist(histos, ax=ax, label=labels, color=colors)
+
+
+# Add the number of entries on top of each bar
+# Get the correct shift in x-axis for each bar
+def calculate_shifts(width, n_bars):
+    half_width = width / 2
+    shift = np.linspace(-half_width, half_width, n_bars, endpoint=False)
+    shift += width / (2 * n_bars)
+    return shift
+
+
+bin_width = 0.8
+shift = calculate_shifts(bin_width, len(histos))
+
+# Loop over the histograms, add on top of each bar the number of entries
+for i, histo in enumerate(histos):
+    for j, value in enumerate(histo.values()):
+        ax.text(
+            j + 0.5 + shift[i],
+            value,
+            int(
+                value
+            ),  # If weighted, f"{height:.1f}" can be used as a better representation of the bin content
+            color="black",
+            ha="center",
+            va="bottom",
+        )
+
+# Set the x-ticks to the middle of the bins and label them
+ax.set_xlim(0, len(categories))
+ax.set_xticks([i + 0.5 for i in range(len(categories))])
+ax.set_xticklabels(categories)
+ax.minorticks_off()
+# Get nice looking y-axis ticks
+ax.set_ylim(top=int(np.max([np.max(histo.values()) for histo in histos]) * 1.5))
+
+ax.set_xlabel("Category")
+ax.set_ylabel("Entries")
+ax.legend()
+
+fig.savefig("1d_side_by_side_with_numbers.svg", bbox_inches="tight")

plothist/examples/advanced/README.rst

@@ -0,0 +1,4 @@
+Other advanced plots
+--------------------
+
+Gallery of images that are used in the other advanced plots section.

plothist/examples/advanced/asymmetry_comparison_advanced.py

@@ -0,0 +1,133 @@
+"""
+Advanced asymmetry comparison
+=============================
+
+Plot the asymmetry between 2 histograms and 2 functions.
+"""
+
+# Get dummy data
+from plothist_utils import get_dummy_data
+
+df = get_dummy_data()
+
+###
+from scipy.stats import norm
+
+from plothist import (
+    add_luminosity,
+    add_text,
+    create_comparison_figure,
+    make_hist,
+    plot_comparison,
+    plot_error_hist,
+    plot_function,
+)
+
+
+# Define some random functions that will be used as Data fit functions
+def f1(x):
+    return 4000 * norm.pdf(x, loc=-0.5, scale=1.6)
+
+
+def f2(x):
+    return 4000 * norm.pdf(x, loc=0.5, scale=1.6)
+
+
+name = "variable_1"
+category = "category"
+
+x1 = df[name][df[category] == 5]
+
+x_range = (-9, 9)
+
+# Create the histograms used as data
+h1 = make_hist(x1 - 2.5, bins=50, range=x_range)
+h2 = make_hist(x1 - 1.5, bins=50, range=x_range)
+
+# Create the figure
+fig, (ax_main, ax_comparison) = create_comparison_figure(
+    gridspec_kw={"height_ratios": [2, 1]}
+)
+
+# Define the marker style
+marker_1 = {
+    "color": "tab:red",
+    "markeredgecolor": "tab:red",
+    "ls": "None",
+    "fmt": "o",
+    "markersize": 5,
+    "label": "$Data_1$",
+}
+
+marker_2 = {
+    "color": "tab:blue",
+    "markerfacecolor": "white",
+    "markeredgecolor": "tab:blue",
+    "ls": "None",
+    "fmt": "o",
+    "markersize": 5,
+    "label": "$Data_2$",
+}
+
+# Plot the data
+plot_error_hist(
+    h1,
+    ax_main,
+    uncertainty_type="symmetrical",
+    density=False,
+    **marker_1,
+)
+
+plot_error_hist(
+    h2,
+    ax_main,
+    uncertainty_type="symmetrical",
+    density=False,
+    **marker_2,
+)
+
+# Plot the functions
+plot_function(f1, x_range, ax_main, color=marker_1["color"], label="Data$_1$ fit")
+plot_function(
+    f2, x_range, ax_main, color=marker_2["color"], linestyle="--", label="Data$_2$ fit"
+)
+
+# Plot the asymmetry comparison between the 2 histograms
+plot_comparison(
+    h1,
+    h2,
+    ax=ax_comparison,
+    h1_label=r"$Data_1$",
+    h2_label=r"$Data_2$",
+    comparison="asymmetry",
+    comparison_ylim=(-1, 1),
+)
+
+
+# Define the asymmetry of the 2 functions
+def asymmetry(x):
+    return (f1(x) - f2(x)) / (f1(x) + f2(x))
+
+
+# Plot the asymmetry of the 2 functions
+plot_function(asymmetry, x_range, ax_comparison, color="black")
+
+ax_main.legend()
+
+fig.align_ylabels()
+
+ax_main.set_xlim(x_range)
+ax_main.set_ylim(ymin=0)
+ax_main.set_ylabel("Entries")
+ax_main.legend()
+
+ax_comparison.set_xlim(x_range)
+ax_comparison.set_xlabel(name)
+
+add_text("Data and fit asymmetry comparison", ax=ax_comparison, x="right")
+
+add_luminosity(
+    collaboration="LMN 3", ax=ax_main, lumi="(1 + 0.3)", preliminary=True, x="left"
+)
+
+fig.savefig("asymmetry_comparison_advanced.svg", bbox_inches="tight")

plothist/examples/advanced/model_examples_flatten2D.py

@@ -0,0 +1,86 @@
+"""
+Flatten 2D histograms
+=====================
+
+This example shows how to flatten and display 2D histograms on one dimension.
+"""
+
+from plothist_utils import get_dummy_data
+
+df = get_dummy_data()
+
+###
+from plothist import (
+    flatten_2d_hist,
+    get_color_palette,
+    make_2d_hist,
+    plot_data_model_comparison,
+    plot_hist,
+)
+
+# Define the histograms
+
+key1 = "variable_1"
+key2 = "variable_2"
+# Bins [-12,0], [0,12] for variable 1,
+# and bins [-12,-5], [-5,0], [0,5], [5,12] for variable 2
+bins = [[-12, 0, 12], [-12, -5, 0, 5, 12]]
+category = "category"
+
+# Define datasets
+
+signal_mask = df[category] == 7
+data_mask = df[category] == 8
+
+background_categories = [0, 1, 2, 3, 4, 5, 6]
+background_categories_labels = [f"c{i}" for i in background_categories]
+background_categories_colors = get_color_palette(
+    "cubehelix", len(background_categories)
+)
+
+background_masks = [df[category] == p for p in background_categories]
+
+# Make histograms
+
+data_hist = make_2d_hist(
+    [df[key][data_mask] for key in [key1, key2]], bins=bins, weights=1
+)
+background_hists = [
+    make_2d_hist([df[key][mask] for key in [key1, key2]], bins=bins, weights=1)
+    for mask in background_masks
+]
+signal_hist = make_2d_hist(
+    [df[key][signal_mask] for key in [key1, key2]], bins=bins, weights=1
+)
+
+# Flatten the 2D histograms
+data_hist = flatten_2d_hist(data_hist)
+background_hists = [flatten_2d_hist(h) for h in background_hists]
+signal_hist = flatten_2d_hist(signal_hist)
+
+# Compare data and stacked histogram
+fig, ax_main, ax_comparison = plot_data_model_comparison(
+    data_hist=data_hist,
+    stacked_components=background_hists,
+    stacked_labels=background_categories_labels,
+    stacked_colors=background_categories_colors,
+    xlabel=rf"({key1} $\times$ {key2}) bin number",
+    ylabel="Entries",
+)
+
+plot_hist(
+    signal_hist,
+    ax=ax_main,
+    color="red",
+    label="Signal",
+    histtype="step",
+)
+
+for ax in [ax_main, ax_comparison]:
+    ax.set_xticks([i + 0.5 for i in range(8)])
+    ax.tick_params(axis="x", which="minor", bottom=False)
+ax_comparison.set_xticklabels([str(i + 1) for i in range(8)])
+
+ax_main.legend(ncol=3, fontsize=10, loc="upper left")
+
+fig.savefig("model_examples_flatten2D.svg", bbox_inches="tight")

plothist/examples/func_1d/README.rst

@@ -0,0 +1,4 @@
+Plot simple functions
+---------------------
+
+Gallery of images that are used in the basic function section.

plothist/examples/func_1d/fct_1d.py

@@ -0,0 +1,27 @@
+"""
+Function
+========
+
+Plot a 1D function with ``plot_function()``.
+"""
+
+###
+from matplotlib import pyplot as plt
+from scipy.stats import norm
+
+from plothist import plot_function
+
+
+# Define the gaussian function of mean=0.5 and std_dev=3
+def f(x):
+    return 1000 * norm.pdf(x, loc=0.5, scale=3)
+
+
+fig, ax = plt.subplots()
+
+plot_function(f, range=(-10, 10), ax=ax)
+
+ax.set_xlabel("x")
+ax.set_ylabel("f(x)")
+
+fig.savefig("fct_1d.svg", bbox_inches="tight")

plothist/examples/func_1d/fct_1d_stacked.py

@@ -0,0 +1,42 @@
+"""
+Stacked functions
+=================
+
+Plot stacked functions using ``plot_function()``.
+"""
+
+from scipy.stats import norm
+
+
+# Define the gaussian function of mean=0.5 and std_dev=3
+def f(x):
+    return 1000 * norm.pdf(x, loc=0.5, scale=3)
+
+
+###
+from matplotlib import pyplot as plt
+
+from plothist import plot_function
+
+
+# Another function
+def g(x):
+    return 1000 * norm.pdf(x, loc=2, scale=3)
+
+
+fig, ax = plt.subplots()
+
+plot_function(
+    [f, g],
+    range=(-10, 10),
+    ax=ax,
+    labels=["f1", "f2"],
+    stacked=True,
+)
+
+ax.set_xlabel("x")
+ax.set_ylabel("f(x)")
+ax.set_ylim(0)
+ax.legend()
+
+fig.savefig("fct_1d_stacked.svg", bbox_inches="tight")

plothist/examples/model_ex/README.rst

@@ -0,0 +1,4 @@
+Model and data
+--------------
+
+Gallery of images that are used in the plotting model and data section.

plothist/examples/model_ex/model_all_comparisons.py

@@ -0,0 +1,103 @@
+"""
+Data/model comparisons
+======================
+
+All supported comparisons between data and model.
+"""
+
+from plothist_utils import get_dummy_data
+
+df = get_dummy_data()
+
+from plothist import get_color_palette, make_hist
+
+# Define the histograms
+
+key = "variable_1"
+range = (-9, 12)
+category = "category"
+
+# Define masks
+signal_mask = df[category] == 7
+data_mask = df[category] == 8
+
+background_categories = [0, 1, 2]
+background_categories_labels = [f"c{i}" for i in background_categories]
+background_categories_colors = get_color_palette(
+    "cubehelix", len(background_categories)
+)
+
+background_masks = [df[category] == p for p in background_categories]
+
+# Make histograms
+data_hist = make_hist(df[key][data_mask], bins=50, range=range, weights=1)
+background_hists = [
+    make_hist(df[key][mask], bins=50, range=range, weights=1)
+    for mask in background_masks
+]
+
+# Optional: scale to data
+background_scaling_factor = data_hist.sum().value / sum(background_hists).sum().value
+background_hists = [background_scaling_factor * h for h in background_hists]
+
+###
+from plothist import (
+    add_text,
+    create_comparison_figure,
+    plot_comparison,
+    plot_data_model_comparison,
+    set_fitting_ylabel_fontsize,
+)
+
+fig, axes = create_comparison_figure(
+    figsize=(6, 13),
+    nrows=6,
+    gridspec_kw={"height_ratios": [3, 1, 1, 1, 1, 1]},
+    hspace=0.3,
+)
+background_sum = sum(background_hists)
+
+plot_data_model_comparison(
+    data_hist=data_hist,
+    stacked_components=background_hists,
+    stacked_labels=background_categories_labels,
+    stacked_colors=background_categories_colors,
+    xlabel="",
+    ylabel="Entries",
+    comparison="ratio",
+    fig=fig,
+    ax_main=axes[0],
+    ax_comparison=axes[1],
+)
+
+add_text(
+    r"Multiple data-model comparisons, $\mathbf{with}$ model uncertainty",
+    ax=axes[0],
+)
+add_text(r'  $\mathbf{→}$ comparison = "ratio"', ax=axes[1], fontsize=13)
+
+for k_comp, comparison in enumerate(
+    ["split_ratio", "pull", "relative_difference", "difference"], start=2
+):
+    ax_comparison = axes[k_comp]
+
+    plot_comparison(
+        data_hist,
+        background_sum,
+        ax=ax_comparison,
+        comparison=comparison,
+        xlabel="",
+        h1_label="Data",
+        h2_label="Pred.",
+        h1_uncertainty_type="asymmetrical",
+    )
+    add_text(
+        rf'  $\mathbf{{→}}$ comparison = "{comparison}"',
+        ax=ax_comparison,
+        fontsize=13,
+    )
+    set_fitting_ylabel_fontsize(ax_comparison)
+
+axes[-1].set_xlabel(key)
+
+fig.savefig("model_all_comparisons.svg", bbox_inches="tight")

plothist/examples/model_ex/model_all_comparisons_no_model_unc.py

@@ -0,0 +1,115 @@
+"""
+Data/model comparisons, no model uncertainty
+============================================
+
+All supported comparisons between data and model, without model uncertainty.
+"""
+
+from plothist_utils import get_dummy_data
+
+df = get_dummy_data()
+
+from plothist import get_color_palette, make_hist
+
+# Define the histograms
+
+key = "variable_1"
+range = (-9, 12)
+category = "category"
+
+# Define masks
+signal_mask = df[category] == 7
+data_mask = df[category] == 8
+
+background_categories = [0, 1, 2]
+background_categories_labels = [f"c{i}" for i in background_categories]
+background_categories_colors = get_color_palette(
+    "cubehelix", len(background_categories)
+)
+
+background_masks = [df[category] == p for p in background_categories]
+
+# Make histograms
+data_hist = make_hist(df[key][data_mask], bins=50, range=range, weights=1)
+background_hists = [
+    make_hist(df[key][mask], bins=50, range=range, weights=1)
+    for mask in background_masks
+]
+
+# Optional: scale to data
+background_scaling_factor = data_hist.sum().value / sum(background_hists).sum().value
+background_hists = [background_scaling_factor * h for h in background_hists]
+
+###
+import numpy as np
+
+from plothist import (
+    add_text,
+    create_comparison_figure,
+    plot_comparison,
+    plot_data_model_comparison,
+    set_fitting_ylabel_fontsize,
+)
+
+fig, axes = create_comparison_figure(
+    figsize=(6, 13),
+    nrows=6,
+    gridspec_kw={"height_ratios": [3, 1, 1, 1, 1, 1]},
+    hspace=0.3,
+)
+background_sum = sum(background_hists)
+
+plot_data_model_comparison(
+    data_hist=data_hist,
+    stacked_components=background_hists,
+    stacked_labels=background_categories_labels,
+    stacked_colors=background_categories_colors,
+    xlabel="",
+    ylabel="Entries",
+    model_uncertainty=False,  # <--
+    comparison="ratio",
+    fig=fig,
+    ax_main=axes[0],
+    ax_comparison=axes[1],
+)
+
+add_text(
+    r"Multiple data-model comparisons, $\mathbf{without}$ model uncertainty",
+    ax=axes[0],
+)
+add_text(r'  $\mathbf{→}$ comparison = "ratio"', ax=axes[1], fontsize=13)
+
+for k_comp, comparison in enumerate(
+    ["split_ratio", "pull", "relative_difference", "difference"], start=2
+):
+    ax_comparison = axes[k_comp]
+
+    # Copy the original histogram and set the uncertainties of the copy to 0.
+    background_sum_copy = background_sum.copy()
+    background_sum_copy[:] = np.c_[
+        background_sum_copy.values(), np.zeros_like(background_sum_copy.values())
+    ]
+
+    plot_comparison(
+        data_hist,
+        background_sum_copy,
+        ax=ax_comparison,
+        comparison=comparison,
+        xlabel="",
+        h1_label="Data",
+        h2_label="Pred.",
+        h1_uncertainty_type="asymmetrical",
+    )
+    if comparison == "pull":
+        # Since the uncertainties of the model are neglected, the pull label is "(Data - Pred.)/sigma_Data"
+        ax_comparison.set_ylabel(r"$\frac{Data-Pred.}{\sigma_{Data}}$")
+    add_text(
+        rf'  $\mathbf{{→}}$ comparison = "{comparison}"',
+        ax=ax_comparison,
+        fontsize=13,
+    )
+    set_fitting_ylabel_fontsize(ax_comparison)
+
+axes[-1].set_xlabel(key)
+
+fig.savefig("model_all_comparisons_no_model_unc.svg", bbox_inches="tight")

plothist/examples/model_ex/model_examples_pull.py

@@ -0,0 +1,56 @@
+"""
+Pull plot
+=========
+
+Compare data and model with pulls.
+"""
+
+from plothist_utils import get_dummy_data
+
+df = get_dummy_data()
+
+from plothist import get_color_palette, make_hist
+
+# Define the histograms
+
+key = "variable_1"
+range = (-9, 12)
+category = "category"
+
+# Define masks
+signal_mask = df[category] == 7
+data_mask = df[category] == 8
+
+background_categories = [0, 1, 2]
+background_categories_labels = [f"c{i}" for i in background_categories]
+background_categories_colors = get_color_palette(
+    "cubehelix", len(background_categories)
+)
+
+background_masks = [df[category] == p for p in background_categories]
+
+# Make histograms
+data_hist = make_hist(df[key][data_mask], bins=50, range=range, weights=1)
+background_hists = [
+    make_hist(df[key][mask], bins=50, range=range, weights=1)
+    for mask in background_masks
+]
+
+# Optional: scale to data
+background_scaling_factor = data_hist.sum().value / sum(background_hists).sum().value
+background_hists = [background_scaling_factor * h for h in background_hists]
+
+###
+from plothist import plot_data_model_comparison
+
+fig, ax_main, ax_comparison = plot_data_model_comparison(
+    data_hist=data_hist,
+    stacked_components=background_hists,
+    stacked_labels=background_categories_labels,
+    stacked_colors=background_categories_colors,
+    xlabel=rf"${key}\,\,[TeV/c^2]$",
+    ylabel="Candidates per 0.42 $TeV/c^2$",
+    comparison="pull",
+)
+
+fig.savefig("model_examples_pull.svg", bbox_inches="tight")