Fused in 5 Minutes
Welcome to Fused! This tutorial will get you up and running in just 5 minutes with three simple examples that demonstrate the core concepts of User Defined Functions (UDFs).
What is a UDF?
A User Defined Function (UDF) is a Python function decorated with @fused.udf
that can be run anywhere - in the cloud, on your local machine, or embedded in web applications. Read more about why UDFs.
Example 1: Hello World UDF
Let's start with the simplest possible UDF:
import fused
@fused.udf
def udf(x: int = 1):
return f"Hello world {x + x}"
fused.run(udf)
What this does:
- Runs on Fused server on a real machine
- The
@fused.udf
decorator makes this function runnable on Fused
Try it:
- Copy this code into Fused Workbench
- Click "Run" to execute it
- You'll see the output: "Hello world 2"
Example 2: Simple Data UDF
Now let's create a UDF that returns structured data:
@fused.udf
def udf(x: int = 5):
import pandas as pd
return pd.DataFrame({"x": range(x)})
What this does:
- Creates a pandas DataFrame with numbers from 0 to x-1
- Returns structured data that can be visualized or processed further
- Default value creates a DataFrame with 5 rows
Try it:
- Paste this code into a new UDF in Workbench
- Run it to see a table with values 0, 1, 2, 3, 4
- Try changing the parameter
x
to see different amounts of data
Example 3: Simple Data File UDF
Finally, let's read real data from a file:
@fused.udf
def udf(x: int = 10):
import pandas as pd
# Read from a public dataset
df = pd.read_parquet("s3://fused-asset/infra/building_msft_us.parquet")
return df.head(x)
What this does:
- Reads a parquet file from cloud storage (S3)
- Returns the first
x
rows of the dataset - This example uses Microsoft building footprint data
Try it:
- Paste this code into Workbench
- Run it to see building data
- Try the File Explorer in Workbench to browse other available datasets
Next Steps
Congratulations! You've just created your first three UDFs. Here's what you can explore next:
- File Explorer: Browse available datasets in the Fused catalog
- Map View: Visualize geospatial data on an interactive map
- Sharing: Share your UDFs with others using shareable URLs
- Advanced Features: Explore caching, async operations, and more complex data transformations
Key Concepts Learned
✅ UDF Decorator: @fused.udf
transforms any Python function into a distributed, scalable function
✅ Type Hints: Parameters should have type hints and default values
✅ Data Returns: UDFs can return strings, DataFrames, or other data structures
✅ Cloud Integration: Easily read from cloud storage like S3
Ready to build more complex UDFs? Check out our Core Concepts and Examples sections!