Interactive dataframes

marimo makes you more productive when working with dataframes.

Display dataframes in a rich, interactive table and chart views
Transform dataframes with filters, groupbys, aggregations, and more, no code required
Select data from tables or charts and get selections back in Python as dataframes

marimo integrates with Pandas and Polars dataframes natively.

For a video overview on interactive dataframes, check out our YouTube tutorial.

Displaying dataframes

marimo lets you page through, search, sort, and filter dataframes, making it extremely easy to get a feel for your data.

marimo brings dataframes to life.

Display dataframes by including them in the last expression of the cell, just like any other object.

/// tab | pandas

import pandas as pd

df = pd.read_json(
    "https://raw.githubusercontent.com/vega/vega-datasets/master/data/cars.json"
)
df

///

/// tab | polars

import polars as pl
import urllib.request

url = "https://raw.githubusercontent.com/vega/vega-datasets/master/data/cars.json"

with urllib.request.urlopen(url) as response:
    df = pl.read_json(response.read())
df

///

/// tab | live example

/// marimo-embed size: large

@app.cell
def __():
    import pandas as pd

    pd.read_json(
    "https://raw.githubusercontent.com/vega/vega-datasets/master/data/cars.json"
    )
    return

///

To opt out of the rich dataframe viewer, use [mo.plain][marimo.plain]:

/// tab | pandas

import pandas as pd
import marimo as mo

df = pd.read_json(
"https://raw.githubusercontent.com/vega/vega-datasets/master/data/cars.json"
)
mo.plain(df)

///

/// tab | polars

import polars as pl
import marimo as mo

df = pl.read_json(
"https://raw.githubusercontent.com/vega/vega-datasets/master/data/cars.json"
)
mo.plain(df)

///

/// tab | live example

/// marimo-embed size: large

@app.cell
def __():
    import pandas as pd

    df = pd.read_json(
    "https://raw.githubusercontent.com/vega/vega-datasets/master/data/cars.json"
    )
    mo.plain(df)
    return

///

Transforming dataframes

No-code transformations

Use [mo.ui.dataframe][marimo.ui.dataframe] to interactively transform a dataframe with a GUI, no coding required. When you're done, you can copy the code that the GUI generated for you and paste it into your notebook.

Build transformations using a GUI

The transformations you apply will turn into code which is accessible via the "code" tab.

Copy the code of the transformation

/// tab | pandas

# Cell 1
import marimo as mo
import pandas as pd

df = pd.DataFrame({"person": ["Alice", "Bob", "Charlie"], "age": [20, 30, 40]})
transformed_df = mo.ui.dataframe(df)
transformed_df

# Cell 2
# transformed_df.value holds the transformed dataframe
transformed_df.value

///

/// tab | polars

# Cell 1
import marimo as mo
import polars as pl

df = pl.DataFrame({"person": ["Alice", "Bob", "Charlie"], "age": [20, 30, 40]})
transformed_df = mo.ui.dataframe(df)
transformed_df

# Cell 2
# transformed_df.value holds the transformed dataframe
transformed_df.value

///

/// tab | live example

/// marimo-embed size: large

@app.cell
def __():
    import pandas as pd

    df = pd.DataFrame({"person": ["Alice", "Bob", "Charlie"], "age": [20, 30, 40]})
    transformed_df = mo.ui.dataframe(df)
    transformed_df
    return

@app.cell
def __():
    transformed_df.value

    return

///

Formatting values

Use format_mapping to format values for display in the dataframe UI. This affects how values appear in the table but does not change the underlying data returned by .value or downloads.

import marimo as mo
import pandas as pd

df = pd.DataFrame(
    {"person": ["Alice", "Bob"], "age": [20, 30], "height_cm": [165.2, 180.4]}
)

def format_height(value: float) -> str:
    return f"{value:.1f} cm"

mo.ui.dataframe(
    df,
    format_mapping={
        "age": "{:d} years".format,
        "height_cm": format_height,
    },
)

Custom filters

Create custom filters with marimo UI elements, like sliders and dropdowns.

/// tab | pandas

# Cell 1 - create a dataframe
df = pd.DataFrame({"person": ["Alice", "Bob", "Charlie"], "age": [20, 30, 40]})

# Cell 2 - create a filter
age_filter = mo.ui.slider(start=0, stop=100, value=50, label="Max age")
age_filter

# Cell 3 - display the transformed dataframe
filtered_df = df[df["age"] < age_filter.value]
mo.ui.table(filtered_df)

///

/// tab | polars

# Cell 1
import marimo as mo
import polars as pl

df = pl.DataFrame({
    "name": ["Alice", "Bob", "Charlie", "David"],
    "age": [25, 30, 35, 40],
    "city": ["New York", "London", "Paris", "Tokyo"]
})

age_filter = mo.ui.slider.from_series(df["age"], label="Max age")
city_filter = mo.ui.dropdown.from_series(df["city"], label="City")

mo.hstack([age_filter, city_filter])

# Cell 2
filtered_df = df.filter((pl.col("age") <= age_filter.value) & (pl.col("city") == city_filter.value))
mo.ui.table(filtered_df)

///

/// tab | live example

/// marimo-embed size: large

@app.cell
def __():
    import pandas as pd

    df = pd.DataFrame({"person": ["Alice", "Bob", "Charlie"], "age": [20, 30, 40]})
    return

@app.cell
def __():
    age_filter = mo.ui.slider(start=0, stop=100, value=50, label="Max age")
    age_filter
    return

@app.cell
def __():
    filtered_df = df[df["age"] < age_filter.value]
    mo.ui.table(filtered_df)
    return

///

Select dataframe rows {#selecting-dataframes}

Display dataframes as interactive, selectable charts using [mo.ui.altair_chart][marimo.ui.altair_chart] or [mo.ui.plotly][marimo.ui.plotly], or as a row-selectable table with [mo.ui.table][marimo.ui.table]. Select points in the chart, or select a table row, and your selection is automatically sent to Python as a subset of the original dataframe.

Select rows in a table, get them back as a dataframe

/// tab | pandas

# Cell 1 - display a dataframe
import marimo as mo
import pandas as pd

df = pd.DataFrame({"a": [1, 2, 3], "b": [4, 5, 6]})
table = mo.ui.table(df, selection="multi")
table

# Cell 2 - display the selection
table.value

///

/// tab | polars

# Cell 1 - display a dataframe
import marimo as mo
import polars as pl

df = pl.DataFrame({"a": [1, 2, 3], "b": [4, 5, 6]})
table = mo.ui.table(df, selection="multi")
table

# Cell 2 - display the selection
table.value

///

/// tab | live example

/// marimo-embed size: large

@app.cell
def __():
    import pandas as pd

    df = pd.DataFrame({"a": [1, 2, 3], "b": [4, 5, 6]})
    table = mo.ui.table(df, selection="multi")
    table
    return

@app.cell
def __():
    table.value
    return

///

Dataframe panels {#dataframe-panels}

Dataframe outputs in marimo come with several panels to help you visualize, explore, and page through your data interactively. These panels are accessible via toggles at the bottom-left of a dataframe output. If you need further control, after opening a panel you can

pin the panel to the side of your editor for persistent access;
toggle focus to automatically display the currently focused dataframe in the panel.

??? note

Toggles are visible when editing notebooks (with `marimo edit ...`) but not when running notebooks as apps (with `marimo run ...`), except for the row viewer which is available in both.

Row viewer panel {#row-viewer-panel}

To inspect individual rows, open the row viewer. This presents a vertical view of the selected row.

Press Space to select/deselect the current row
Use arrow keys (← →) to navigate between rows
Click on any row in the dataframe to view its data in the panel

Column explorer panel {#column-explorer-panel}

To explore your data, open the column explorer where you can find summary statistics and charts for each column. Click the + button to add the chart code to a new cell.

This requires the altair package to be installed. For large dataframes, vegafusion is also needed to render charts. To use the generated Python code, enable vegafusion in your notebook:

import altair

altair.data_transformers.enable("vegafusion")

Chart builder {#chart-builder}

The chart builder toggle lets you rapidly develop charts using a GUI, while also generating Python code to insert in your notebook. Refer to the chart builder guide for more details.

Preferences {#preferences}

When you run a SQL cell in marimo, you can get the output returned as a dataframe. If you have a preference for a specific dataframe library as a default you can configure the "default SQL output" in the user settings by going to the "Runtime" tab.

Configure the default SQL output

Alternatively you can also use the marimo configuration file to configure the default SQL output.

[runtime]
default_sql_output = "native"

Example notebook {#example-notebook}

For a comprehensive example of using Polars with marimo, check out our Polars example notebook.

Run it with:

marimo edit https://raw.githubusercontent.com/marimo-team/marimo/main/examples/third_party/polars/polars_example.py

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Interactive dataframes

Displaying dataframes

Transforming dataframes

No-code transformations

Formatting values

Custom filters

Select dataframe rows {#selecting-dataframes}

Dataframe panels {#dataframe-panels}

Row viewer panel {#row-viewer-panel}

Column explorer panel {#column-explorer-panel}

Chart builder {#chart-builder}

Preferences {#preferences}

Example notebook {#example-notebook}

FilesExpand file tree

dataframes.md

Latest commit

History

dataframes.md

File metadata and controls

Interactive dataframes

Displaying dataframes

Transforming dataframes

No-code transformations

Formatting values

Custom filters

Select dataframe rows {#selecting-dataframes}

Dataframe panels {#dataframe-panels}

Row viewer panel {#row-viewer-panel}

Column explorer panel {#column-explorer-panel}

Chart builder {#chart-builder}

Preferences {#preferences}

Example notebook {#example-notebook}