Learn how to define and use custom parameters while downloading assets or
using dataset in Compute-to-data environment
---
# User defined parameters
### Overview
Ocean Protocol allows dataset buyers to provide custom parameters that can be used to fetch the downloaded data in a specific format, download a different type of data or pass some additional input to the algorithms in the Compute-to-Data job.
There 2 are types of parameters that asset publishers can support:
The dataset publisher can support these parameters to allow filtering or querying of the published dataset. The additional parameters that facilitate this are called `User defined parameters`. The Provider combines the original asset URL and the entered parameter values into a new URL and then streams the response from the modified URL back to the buyer.
#### Use case for user defined parameters
For example, if the publisher has published an URL `https://example.com` which serves large size historical weather data from all over the world, the publisher could allow buyers to filter the data based on location, type of data, etc. It is possible to do this using user defined parameters.
Suppose the publisher defines the following 2 parameters:
Suppose the buyer wants to download the temperature data in the region code `XYZ`. While downloading the data, the buyer enters the desired parameter values using ocean.py.
The provider will decrypt the URL from the DDO published on-chain, construct a new URL with the additional parameters and finally stream data to the buyer.
Internally, the new URL will be of the format `https://example.com/?location=XYZ&type=temperature`. The server hosting the data has to read these parameters and serve the appropriate data.
The following steps will specify how the publisher can support additional parameters.
#### Step 1: Create a service
The below python script exposes a REST endpoint that takes two parameters, namely: `location` and `type`. Let's assume that the dataset publisher hosts the service at domain `example.com` along with HTTPS support. The publisher must ensure that the URL is accessible to Provider.
The code snippet is only for demo purposes and not for production use.
```python
from flask import Flask, request
def get_data(data_type: str, location: str):
'''
Add some business logic here to get
the required data with given parameters
'''
return {}
@app.route('/', methods=['GET'])
def serve_content():
args = request.args
data_type = args.get('type')
location = args.get('location')
result = get_data(data_type, location)
return result
```
#### Step 2: Publish dataset asset with compute service
For a complete tutorial on publishing asset using Ocean Marketplace read [our guide on publishing with Ocean Market](../../using-ocean-market/marketplace-publish-data-asset.md).
### Publish an algorithm that uses custom parameters
#### Use case for algorithm custom parameters
For example, if the algorithm publisher has published an URL `https://example.org` which serves python script to analyze historical weather data published in the previous section. If the algorithm publisher wants buyers to specify the number of iterations the algorithm must perform over the data, it is possible to do so using algorithm custom parameters.
Suppose the algorithm publisher defines a parameter called `iterations` and expects the buyer to give this input before running the algorithm in a Compute-to-Data environment. The buyer can enter the desired parameter value using ocean.py or ocean.js.
The provider passes the entered parameters and saves them in a specific path in the Compute-to-Data environment. The algorithm can later read this value and perform required computations.
The following steps will specify how the algorithm publisher can support additional algorithm custom parameters.
#### Step 1: Create an algorithm
The code snippet is only for demo purposes and not for production use.
For a complete tutorial on publishing asset using Ocean Marketplace read [our guide on publishing with Ocean Market](../../using-ocean-market/marketplace-publish-data-asset.md).
### Starting compute job with custom parameters
In this example, the buyer wants to run the algorithm with certain parameters on a selected dataset. The code snippet below shows how the buyer can start the compute job with custom parameter values. Before embarking on this tutorial you should familiarize yourself with how to:
- Search for a dataset using [Ocean market](https://market.oceanprotocol.com/) or [Aquarius API](../../api-references/aquarius-rest-api.md)
- [Allow an algorithm to run on the dataset](https://github.com/oceanprotocol/ocean.py/blob/6eb068df338abc7376430cc5ba7fe2d381508328/READMEs/c2d-flow.md#5-alice-allows-the-algorithm-for-c2d-for-that-data-asset)
- Buy datatokens using [Ocean market](https://market.oceanprotocol.com/) or [ocean.py](https://github.com/oceanprotocol/ocean.py)
For configuring ocean.py/ocean.js, please refer this [guide](../ocean-libraries/configuration.md). Copy the below code snippet to a file locally after completing required configurations and execute the script.
