`Round`#

Description#

Round each element to the nearest integer using round-half-to-even (banker's rounding). Halves go to the nearest even integer rather than always away from zero, which matches numpy.round and Python's built-in round.

Equation:

\[ y[i] = \operatorname{round}(x[i]) \]

with halves resolved to even.

Parameters: Round takes no parameters.

NaN handling: NaN values pass through unchanged.

Note: This is not the same as the C++ std::round (half-away-from-zero). For example Round(0.5) == 0, Round(1.5) == 2, Round(2.5) == 2.

NaN handling#

Policy: ignore. A NaN in any input at index t causes the function to skip that step: output at t is NaN and internal state is unchanged. Subsequent finite samples are processed as if step t had not occurred.

Examples#

Usage example#

import numpy as np
import plotly.graph_objects as go
from plotly.subplots import make_subplots
from screamer import Round

np.random.seed(0)
data = np.random.normal(size=80) * 1.5
transformed = Round()(data)

fig = make_subplots(rows=2, cols=1, shared_xaxes=True,
                    row_heights=[0.5, 0.5], vertical_spacing=0.1)
fig.add_trace(go.Scatter(y=data, mode="lines+markers",
                         name="input"), row=1, col=1)
fig.add_trace(go.Scatter(y=transformed, mode="lines+markers",
                         name="Round",
                         line=dict(color="red")), row=2, col=1)
fig.update_layout(
    title="Round transformation (banker's rounding)",
    margin=dict(l=20, r=20, t=60, b=20),
    legend=dict(orientation="h", yanchor="bottom", y=1.02,
                xanchor="right", x=1),
)
fig.update_yaxes(title_text="input", row=1, col=1)
fig.update_yaxes(title_text="Round(input)", row=2, col=1)
fig.show()

Reference#

Equivalent to numpy.round and Python's built-in round.

Round#