A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.
Python artificial-intelligencedagdata-sciencedata-visualizationdataflowdeveloper-toolsmachine-learningnotebookspipelinepythonreactivesqlweb-app
# /// script
# requires-python = ">=3.9"
# dependencies = [
# "marimo",
# "pandas==2.2.3",
# ]
# ///
import marimo
__generated_with = "0.19.7"
app = marimo.App(width="medium")
@app.cell(hide_code=True)
def _(mo):
mo.md(r"""
# Filterable DataFrame
""")
return
@app.cell
def _(pd):
# Read the csv
df = pd.read_json(data_url("cars.json"))
return (df,)
@app.cell
def _(df):
# Create options for select widgets
manufacturer_options = df["Name"].str.split().str[0].unique()
manufacturer_options.sort()
cylinder_options = df["Cylinders"].unique().astype(str)
cylinder_options.sort()
return cylinder_options, manufacturer_options
@app.cell
def _(cylinder_options, df, manufacturer_options, mo):
# Create the filters
manufacturer = mo.ui.dropdown(manufacturer_options, label="Manufacturer")
cylinders = mo.ui.dropdown(cylinder_options, label="Cylinders")
horse_power = mo.ui.range_slider.from_series(
df["Horsepower"],
show_value=True,
)
mo.hstack([manufacturer, horse_power, cylinders], gap=3).left()
return cylinders, horse_power, manufacturer
@app.cell
def _(df, filter_df):
filter_df(df)
return
@app.cell
def _(cylinders, horse_power, manufacturer):
def filter_df(df):
filtered_df = df
if manufacturer.value:
filtered_df = filtered_df[
filtered_df["Name"].str.contains(manufacturer.value, case=False)
]
if cylinders.value:
filtered_df = filtered_df[filtered_df["Cylinders"] == cylinders.value]
if horse_power.value:
left, right = horse_power.value
filtered_df = filtered_df[
(filtered_df["Horsepower"] >= left)
& (filtered_df["Horsepower"] <= right)
]
return filtered_df
return (filter_df,)
@app.function
def data_url(file):
return f"https://cdn.jsdelivr.net/npm/[email protected]/data/{file}"
@app.cell
def _():
import marimo as mo
import pandas as pd
return mo, pd
if __name__ == "__main__":
app.run()