---
title: Downsample data with notebooks
description: Downsample data to summarize data at specified time intervals and reduce the overall disk usage as data collects over time. Learn how to create a notebook that downsamples data.
url: https://docs.influxdata.com/influxdb/cloud/tools/notebooks/downsample/
estimated_tokens: 1216
product: InfluxDB Cloud (TSM)
version: cloud
---

# Downsample data with notebooks

Create a notebook to downsample data. Downsampling aggregates or summarizes data within specified time intervals, reducing the overall disk usage as data collects over time.

The following example creates a notebook that queries **Coinbase bitcoin price sample data** from the last hour, downsamples the data into ten minute summaries, and then writes the downsampled data to an InfluxDB bucket.

1. If you do not have an existing bucket to write the downsampled data to, [create a new bucket](/influxdb/cloud/admin/buckets/create-bucket/).
    
2. [Create a new notebook](/influxdb/cloud/tools/notebooks/create-notebook/).
    
3. Select **Past 1h** from the time range drop-down list at the top of your notebook.
    
4. In the **Build a Query** cell:
    
    1. In the **FROM** column under **Sample**, select **Coinbase bitcoin price**.
    2. In the next **FILTER** column, select **\_measurement** from the drop-down list and select the **coindesk** measurement in the list of measurements.
    3. In the next **FILTER** column, select **\_field** from the drop-down list, and select the **price** field from the list of fields.
    4. In the next **FILTER** column, select **code** from the drop-down list, and select a currency code.
5. Click after your **Build a Query** cell to add a new cell and select **Flux Script**.
    
6. In the Flux script cell:
    
    1. Use `__PREVIOUS_RESULT__` to load the output of the previous notebook cell into the Flux script.
        
    2. Use [`aggregateWindow()`](/flux/v0/stdlib/universe/aggregatewindow/) to window data into ten minute intervals and return the average of each interval. Specify the following parameters:
        
        -   **every**: Window interval *(should be less than or equal to the duration of the queried time range)*. For this example, use `10m`.
        -   **fn**: [Aggregate](/flux/v0/function-types/#aggregates) or [selector](/flux/v0/function-types/#selectors) function to apply to each window. For this example, use `mean`.
    3. Use [`to()`](/flux/v0/stdlib/influxdata/influxdb/to/) to write the downsampled data back to an InfluxDB bucket.
        
    
    ```js
    __PREVIOUS_RESULT__
        |> aggregateWindow(every: 10m, fn: mean)
        |> to(bucket: "example-bucket")
    ```
    
7. *(Optional)* Click and select **Note** to add a note to describe your notebook, for example, “Downsample Coinbase bitcoin prices into hourly averages.”
    
8. Click **Run** to run the notebook and write the downsampled data to your bucket.
    

## Continuously run a notebook

To continuously run your notebook, export the notebook as a task:

1. Click to add a new cell, and then select **Task**.
    
2. Provide the following:
    
    -   **Every**: Interval that the task should run at.
    -   **Offset**: *(Optional)* Time to wait after the defined interval to execute the task. This allows the task to capture late-arriving data.
3. Click **Export as Task**.
    

[notebooks](/influxdb/cloud/tags/notebooks/)