hvPlot

High-level plotting API providing interactive visualizations for pandas, xarray, and other data structures

Interactive Visualization & Exploration Intermediate Recommended Tool Assess Provisional Teaching
Download Documentation

Our View

Assess Provisional Teaching

Assess whether this fits the group’s research and teaching workflows. hvPlot makes creating interactive visualizations as easy as using pandas' built-in plotting, but with interactivity out of the box. It's perfect for exploratory data analysis, allowing you to zoom, pan, and hover over data points without writing complex plotting code.

Provisional positions are editorial proposals awaiting group review.

Teaching — Recorded in topic offerings; consult the dated event links for availability.

1 topics with recorded teaching · 1 past offerings · 0 upcoming offerings

Related themes: Scientific Computing

Related topics: Automating Research Workflows

Common use cases

  • Interactive exploratory data analysis
  • Quick visualization of pandas DataFrames
  • Time series data exploration
  • Multi-dimensional data with xarray

Getting started

hvPlot is a high-level plotting API that provides a familiar interface similar to pandas .plot() but generates interactive Bokeh plots. It works with pandas DataFrames, xarray, and many other data structures.

Why hvPlot?

  • Interactive by Default: Zoom, pan, hover without extra code
  • Familiar API: If you know pandas plotting, you know hvPlot
  • Flexible Output: Bokeh, Matplotlib, or Plotly backends
  • Dashboard Ready: Integrates with Panel for dashboards
  • Multi-Format: Works with pandas, xarray, GeoPandas, NetworkX

Basic Usage

Simple Interactive Plots

import pandas as pd
import hvplot.pandas  # Adds .hvplot accessor

# Load data
df = pd.read_csv('data.csv')

# Interactive line plot (just like .plot() but interactive)
df.hvplot.line(x='time', y='value', width=800, height=400)

# Interactive scatter plot with hover
df.hvplot.scatter(x='age', y='score', c='group', hover_cols=['name'])

Comparison with Pandas

# Pandas (static)
df.plot(x='time', y='value')

# hvPlot (interactive)
df.hvplot(x='time', y='value')

Common Plot Types

Line Plots

# Single line
df.hvplot.line(x='time', y='voltage')

# Multiple lines
df.hvplot.line(x='time', y=['voltage', 'current'])

# With customization
df.hvplot.line(
    x='time',
    y='signal',
    title='LFP Recording',
    xlabel='Time (s)',
    ylabel='Voltage (mV)',
    width=900,
    height=400,
    color='blue'
)

Scatter Plots

# Basic scatter
df.hvplot.scatter(x='x_coord', y='y_coord')

# Color by category
df.hvplot.scatter(x='x', y='y', c='category', cmap='Category10')

# Size by value
df.hvplot.scatter(x='x', y='y', s='size', scale=0.5)

# Custom hover information
df.hvplot.scatter(
    x='x', y='y',
    hover_cols=['name', 'value', 'group']
)

Histograms and Distributions

# Histogram
df.hvplot.hist('reaction_time', bins=30)

# Kernel density estimate
df.hvplot.kde('reaction_time')

# Distribution by group
df.hvplot.hist('score', by='condition', alpha=0.5)

Heatmaps

# Correlation matrix
corr_matrix = df.corr()
corr_matrix.hvplot.heatmap(cmap='RdBu_r', clim=(-1, 1))

# Time-frequency representation
spectrogram_df.hvplot.heatmap(
    x='time', y='frequency', C='power',
    cmap='viridis', colorbar=True
)

Working with Time Series

import pandas as pd
import hvplot.pandas

# Time series with datetime index
df.index = pd.to_datetime(df.index)

# Plot with interactive time selection
df.hvplot.line(x='time', y='value', xlabel='Time', ylabel='Signal')

# Multiple channels
channels = ['Ch1', 'Ch2', 'Ch3', 'Ch4']
df.hvplot.line(x='time', y=channels, ylabel='LFP (mV)')

xarray Integration

import xarray as xr
import hvplot.xarray

# Load multi-dimensional data
ds = xr.open_dataset('recording.nc')

# Interactive 2D plot with sliders for other dimensions
ds['lfp'].hvplot.image(x='time', y='channel', cmap='RdBu_r')

# Line plot with interactive dimension selection
ds['power'].hvplot.line(x='frequency', groupby='channel')

# Quadmesh for irregular grids
ds.hvplot.quadmesh(x='time', y='frequency', z='power', cmap='viridis')

Combining Multiple Plots

Overlays

# Overlay plots using *
plot1 = df1.hvplot.line(x='time', y='signal1', label='Control')
plot2 = df2.hvplot.line(x='time', y='signal2', label='Treatment')
plot1 * plot2

Layouts

# Side-by-side using +
plot1 + plot2

# Vertical layout
plot1 + plot2

# Complex layouts
(plot1 + plot2) + (plot3 + plot4)

Advanced Features

Interactive Widgets

# GroupBy for interactive dimension selection
df.hvplot.line(x='time', y='signal', groupby='trial')

# This creates a widget to select different trials

Custom Styling

df.hvplot.scatter(
    x='x', y='y',
    color='blue',
    size=50,
    alpha=0.6,
    title='My Plot',
    fontsize={'title': 16, 'labels': 14, 'xticks': 10, 'yticks': 10},
    grid=True
)

Geographic Data

import hvplot.pandas

# Plot with map tiles
gdf.hvplot.points(
    x='lon', y='lat',
    geo=True,
    tiles='OSM',
    color='red',
    size=50
)

Practical Neuroscience Examples

Calcium Imaging Traces

# Interactive calcium trace explorer
traces_df.hvplot.line(
    x='time',
    y='dF_F',
    groupby='roi',  # Widget to select ROIs
    xlabel='Time (s)',
    ylabel='ΔF/F',
    width=900,
    height=400
)

Spike Raster Plot

# Raster plot with interactive hover
spikes_df.hvplot.scatter(
    x='time',
    y='unit_id',
    hover_cols=['unit_id', 'time', 'trial'],
    s=1,
    color='black',
    ylabel='Unit',
    xlabel='Time (s)'
)

LFP Power Spectra

# Compare power spectra across conditions
power_df.hvplot.line(
    x='frequency',
    y='power',
    by='condition',
    logy=True,
    xlabel='Frequency (Hz)',
    ylabel='Power',
    legend='top_right'
)

Explorer Functionality

# Create an interactive explorer with all plot types
from hvplot import explorer

explorer = df.hvplot.explorer()
explorer  # Interactive GUI to try different plot types

Integration with Panel

hvPlot works seamlessly with Panel for dashboards:

import panel as pn

# Create plot
plot = df.hvplot.line(x='time', y='value')

# Create dashboard
dashboard = pn.Column(
    '# My Dashboard',
    plot
)

dashboard.servable()

Backends

import hvplot.pandas

# Use Bokeh (default)
hvplot.extension('bokeh')

# Use Matplotlib for static plots
hvplot.extension('matplotlib')

# Use Plotly
hvplot.extension('plotly')

Installation

pixi add hvplot
# or
conda install -c conda-forge hvplot
# or
pip install hvplot

Best Practices

  • Use descriptive labels and titles
  • Include relevant columns in hover_cols for context
  • Set appropriate width and height for readability
  • Use groupby for exploring multi-dimensional data
  • Combine with Panel for interactive dashboards
  • Choose colormaps that are colorblind-friendly
  • Cache large plots for better performance
  • Use .opts() for fine-grained control when needed