Time Series Plotting Examples

Basic Time Series Plot

import numpy as np
from mdu.plotly.time_series import plot_ts
from mdu.plotly.template import set_template

set_template()

# Generate sample data
data = np.random.randn(1000, 2)

# Create and display plot
fig = plot_ts(data, names=['Signal A', 'Signal B'])
fig.show()

Custom X-Axis

# Time-based x-axis
time = np.linspace(0, 10, 1000)
data = np.sin(time) + np.random.randn(1000) * 0.1

fig = plot_ts(data, x=time)
fig.update_layout(xaxis_title="Time (s)", yaxis_title="Amplitude")
fig.show()

Large Dataset with Resampling

Resampler and Static HTML

The plotly-resampler functionality requires an interactive Python session to dynamically resample data during pan/zoom operations. It does not work with static HTML exports (like this documentation page). The examples below demonstrate the API usage, but interactive resampling will only work in Jupyter notebooks or live Python sessions.

from mdu.plotly.time_series import plot_ts

# Generate large dataset
large_data = np.random.randn(1_000_000, 1)

# Use resampler for performance
fig = plot_ts(large_data, names=['Large Signal'])
fig.show()

The plot will downsample automatically during pan/zoom operations in interactive sessions.

Multiple Channels

# Simulate multi-channel recording
n_samples = 5000
n_channels = 4
data = np.random.randn(n_samples, n_channels)

# Add some structure
for i in range(n_channels):
    data[:, i] += np.sin(2 * np.pi * (i+1) * np.linspace(0, 10, n_samples))

# Create names
channel_names = [f'Channel {i+1}' for i in range(n_channels)]

fig = plot_ts(data, names=channel_names, use_resampler=False)
fig.update_layout(height=800)
fig.show()

From Example File

Note

The following example demonstrates resampler usage. Remember that resampling only works in interactive sessions, not in static HTML.

The repository includes examples/resampler_example.py:

import numpy as np
from mdu.plotly.time_series import plot_ts

# Generate synthetic multi-channel data
n_samples = 100_000
n_channels = 3

# Create time vector
time = np.linspace(0, 100, n_samples)

# Generate data with different frequencies
data = np.zeros((n_samples, n_channels))
for i in range(n_channels):
    freq = (i + 1) * 0.5
    data[:, i] = np.sin(2 * np.pi * freq * time) + np.random.randn(n_samples) * 0.1

# Plot with resampler (automatically enabled for large data)
fig = plot_ts(
    data,
    x=time,
    names=['Channel 1 (0.5 Hz)', 'Channel 2 (1.0 Hz)', 'Channel 3 (1.5 Hz)'],
    use_resampler=True
)

fig.update_layout(
    title="Multi-Channel Time Series with Resampler",
    xaxis_title="Time (s)",
    height=600
)

fig.show()

Performance Tips

When to Use Resampler

• Use resampler: Datasets > 100k points
• Don’t use resampler: Small datasets, static exports

Memory Considerations

# For very large datasets, process in chunks
chunk_size = 100_000
large_array = np.random.randn(1_000_000, 1)

# Plot first chunk
fig = plot_ts(large_array[:chunk_size], use_resampler=True)

Customization

# Combine with Plotly customization
fig = plot_ts(data, use_resampler=False)

fig.update_traces(
    line=dict(width=1),
    marker=dict(size=3)
)

fig.update_layout(
    template='plotly_dark',
    font=dict(size=14)
)

fig.show()