Histogram

from dataviz import histogram
from dataviz import save
from dataviz.core.colors import REBECCAPURPLE
from dataviz import Theme

def main() raises:
    # The shared illustrative checkout latency sample is intentionally
    # unsorted and right-skewed, so binning reveals its slow tail.
    var latency_ms: List[Int] = [
        84, 72, 91, 68, 75, 88, 79, 73, 96, 82,
        77, 69, 85, 74, 101, 80, 71, 93, 76, 87,
        83, 70, 78, 89, 81, 95, 67, 86, 72, 90,
        110, 124, 138, 156, 205, 98, 105, 118, 74, 82,
    ]

    var c = histogram(
        latency_ms,
        bins=10,
        title="Illustrative Checkout API Latency",
        theme=Theme(mark_color=REBECCAPURPLE),
        x_title="Latency (ms)",
        y_title="Requests",
    )
    save(c, "docs/src/examples/out_histogram.svg")

from dataviz import save_facets
from dataviz import histogram
from dataviz.core.colors import CRIMSON, STEELBLUE
from dataviz.binned.histogram import HistStat, shared_bin_edges
from dataviz import Plot
from dataviz import Theme

def main() raises:
    # Two production lines, same part, measured in microns off
    # nominal. Line B runs wider AND is shifted high; the only way
    # to see both facts at once is for the two histograms to use
    # the same intervals, so a bar in one covers exactly the bar
    # beside it.
    var line_a: List[Float64] = [
        -3.1, -2.4, -2.0, -1.6, -1.5, -1.1, -0.9, -0.8, -0.6, -0.4,
        -0.3, -0.2, 0.0, 0.1, 0.2, 0.4, 0.5, 0.7, 0.9, 1.2,
        1.4, 1.8, 2.2, 2.9,
    ]
    var line_b: List[Float64] = [
        -2.2, -1.3, -0.5, 0.1, 0.4, 0.8, 1.0, 1.3, 1.5, 1.7,
        1.9, 2.0, 2.2, 2.4, 2.6, 2.8, 3.1, 3.4, 3.8, 4.3,
        4.9, 5.6, 6.4, 7.5,
    ]

    # One call, both samples: the pooled min and max in 12 equal
    # intervals. Binning each sample on its own range would give
    # the two charts different bars.
    var groups: List[List[Float64]] = [line_a.copy(), line_b.copy()]
    var edges = shared_bin_edges(groups, bins=12)

    # PROBABILITY, not COUNT, so the two panels are comparable
    # even when the samples are not the same size.
    var a = histogram(
        line_a,
        edges=edges,
        stat=HistStat.PROBABILITY,
        theme=Theme(mark_color=STEELBLUE, show_gridlines=False),
        width=380,
        height=320,
        title="Line A",
        x_title="Microns off nominal",
        y_title="Share of parts",
    )
    var b = histogram(
        line_b,
        edges=edges,
        stat=HistStat.PROBABILITY,
        theme=Theme(mark_color=CRIMSON, show_gridlines=False),
        width=380,
        height=320,
        title="Line B",
        x_title="Microns off nominal",
    )

    # Two panels rather than one overlay: render_layers() refuses
    # the explicit x-domain a histogram pins. The panels
    # still line up bar for bar, because the edges -- and so both
    # x-domains -- are the same list. Their y-axes are separate,
    # though, so read the bar positions across panels and
    # the heights within one.
    save_facets(
        a,
        b,
        cols=2,
        path="docs/src/examples/out_histogram_shared.svg",
        shared_y_scale=True,
    )

from dataviz import save_facets
from dataviz import histogram
from dataviz.core.colors import DARKORANGE, STEELBLUE
from dataviz.binned.histogram import BinRule
from dataviz import Plot
from dataviz import Theme

def main() raises:
    # Fourteen months of rainfall. A small sample is where a
    # fixed bin count does the most damage: ten bins over
    # fourteen observations leaves two of them empty and makes
    # four more a single observation tall, so the chart shows
    # the gaps between the readings rather than the shape of
    # the distribution.
    var rainfall: List[Float64] = [
        12.4, 15.1, 15.8, 17.2, 18.0, 18.3, 19.1,
        19.6, 20.4, 21.7, 23.0, 25.6, 28.9, 34.2,
    ]

    # The default: a count nobody chose for this sample.
    var fixed = histogram(
        rainfall,
        bins=10,
        theme=Theme(mark_color=STEELBLUE, show_gridlines=False),
        width=380,
        height=320,
        title="bins=10",
        subtitle="Two empty bins, four one deep",
        x_title="Monthly rainfall (mm)",
        y_title="Months",
    )

    # BinRule.AUTO reads the sample and answers 5 -- enough to
    # show the right skew, few enough that every bar stands on
    # more than one reading. Swap in SQRT, STURGES, RICE, SCOTT
    # or FREEDMAN_DIACONIS to see what each rule makes of the
    # same numbers.
    var chosen = histogram(
        rainfall,
        bins=BinRule.AUTO,
        theme=Theme(mark_color=DARKORANGE, show_gridlines=False),
        width=380,
        height=320,
        title="bins=BinRule.AUTO",
        subtitle="Five bins, chosen from the data",
        x_title="Monthly rainfall (mm)",
    )

    save_facets(fixed, chosen, cols=2, path="docs/src/examples/out_histogram_auto.svg")

from dataviz import histogram, kdeplot, save_layers
from dataviz import Theme
from dataviz.core.colors import STEELBLUE, CRIMSON
from dataviz.binned.histogram import HistStat
from dataviz import Plot

def main() raises:
    # Reaction times from a task with a slow tail. The bars and
    # the curve share one numeric x-axis, and HistStat.DENSITY
    # puts the bars on the scale the density curve is on, so
    # the two are directly comparable: where the curve runs
    # above a bar the kernel is smoothing over a gap, where it
    # runs below, a spike.
    var seed = 2024
    var reaction_ms = List[Float64]()
    for _ in range(400):
        var s = 0.0
        for _ in range(3):
            seed = (seed * 1103515245 + 12345) % 2147483648
            s += Float64(seed % 10000) / 10000.0
        # Sum of three uniforms is bell-shaped; cubing the last
        # draw skews a share of them slow.
        seed = (seed * 1103515245 + 12345) % 2147483648
        var u = Float64(seed % 10000) / 10000.0
        reaction_ms.append(320.0 + 60.0 * s + 400.0 * u * u * u)
    var bars = histogram(
        reaction_ms,
        bins=24,
        stat=HistStat.DENSITY,
        theme=Theme(mark_color=STEELBLUE, show_gridlines=False),
        title="Illustrative Reaction Times with a Density Curve",
        x_title="Reaction time (ms)",
        y_title="Density",
    )
    var curve = kdeplot(
        reaction_ms,
        theme=Theme(mark_color=CRIMSON, show_gridlines=False),
    )
    save_layers(bars, curve, path="docs/src/examples/out_histogram_density.svg")

from dataviz import histogram, save
from dataviz import Theme
from dataviz.core.colors import DARKORANGE

def main() raises:
    # The shape a marginal histogram takes beside a joint plot's
    # y-axis: bins up the left edge, counts running right. The
    # same latency sample as the first example.
    var latency_ms: List[Int] = [
        84, 72, 91, 68, 75, 88, 79, 73, 96, 82,
        77, 69, 85, 74, 101, 80, 71, 93, 76, 87,
        83, 70, 78, 89, 81, 95, 67, 86, 72, 90,
        110, 124, 138, 156, 205, 98, 105, 118, 74, 82,
    ]
    var c = histogram(
        latency_ms,
        bins=10,
        horizontal=True,
        theme=Theme(mark_color=DARKORANGE),
        title="Illustrative Checkout API Latency, Sideways",
        x_title="Requests",
        y_title="Latency (ms)",
        width=420,
        height=420,
    )
    save(c, "docs/src/examples/out_histogram_horizontal.svg")