PyGWalker Streamlit API Guide

This guide explains how to integrate PyGWalker with Streamlit applications, providing interactive data visualization capabilities.

StreamlitRenderer Class

The StreamlitRenderer class is the main interface for using PyGWalker in Streamlit apps.

from pygwalker.api.streamlit import StreamlitRenderer
 
renderer = StreamlitRenderer(df, spec="./gw_config.json")

Key Parameters

Parameter	Type	Default	Description
dataset	Union[DataFrame, Connector]	-	Input data source
spec	str	""	Chart configuration data (ID, JSON, or URL)
spec_io_mode	Literal["r", "rw"]	"r"	Spec I/O mode: "r" (read) or "rw" (read/write)
kernel_computation	bool	None	Enable high-performance kernel computation
appearance	Literal['media', 'light', 'dark']	'media'	Theme setting
default_tab	Literal["data", "vis"]	"vis"	Default tab to show

Main Methods

1. explorer()

Displays the full GraphicWalker UI.

renderer.explorer(width=None, height=1000, scrolling=False, default_tab="vis")

2. chart(index)

Renders a specific chart.

renderer.chart(index, width=None, height=1000, scrolling=False)

3. viewer()

Displays a view-only version of the charts.

renderer.viewer(width=None, height=1000, scrolling=False)

Best Practices

Caching: Use Streamlit's caching to prevent memory issues:

@st.cache_resource
def get_pyg_renderer():
    df = pd.read_csv("data.csv")
    return StreamlitRenderer(df, spec="./gw_config.json")

Large Datasets: Enable kernel_computation=True for better performance with large datasets.
Layout: Use Streamlit's layout options (e.g., st.set_page_config(layout="wide")) for optimal display.
Tabs: Organize different PyGWalker views using Streamlit tabs for a cleaner interface.

Example: Streamlit App with PyGWalker

import streamlit as st
from pygwalker.api.streamlit import StreamlitRenderer
import pandas as pd
 
st.set_page_config(page_title="PyGWalker in Streamlit", layout="wide")
st.title("PyGWalker in Streamlit Demo")
 
@st.cache_resource
def get_pyg_renderer() -> StreamlitRenderer:
    df = pd.read_csv("data.csv")
    return StreamlitRenderer(df, spec="./gw_config.json", spec_io_mode="rw")
 
renderer = get_pyg_renderer()
 
tab1, tab2, tab3 = st.tabs(["Explorer", "Data Profiling", "Charts"])
 
with tab1:
    renderer.explorer()
 
with tab2:
    renderer.explorer(default_tab="data")
 
with tab3:
    st.subheader("Registered per Weekday")
    renderer.chart(0)
    st.subheader("Registered per Day")
    renderer.chart(1)

Advanced: Building a Tableau/PowerBI Alternative

To create an open-source alternative to Tableau/PowerBI that can handle large datasets:

Use PyGWalker for interactive visualizations
Leverage Streamlit for the web application framework
Integrate with Snowflake for handling large-scale data

This combination provides a powerful, scalable solution for data visualization and analysis.

For more details and advanced usage, refer to the PyGWalker documentation (opens in a new tab).

Related Q&A

How to build a online open source alternative to Tableau/PowerBI which can handle large amount of data?

Answer: You can use pygwalker + streamlit + snowflake to build a online open source alternative to Tableau/PowerBI which can handle large amount of data.

Use in Jupyter Gradio Component