Create, Configure a Scoring Algorithm
Everything you need to know about how to create and set up a pre-trained AI Scoring Algorithm.
The HrFlow.ai Portal offers multiple pre-trained Scoring Algorithms segmented by scopes such as:
- Type of jobs
- Type of profiles
- Sector
- industry.
Prerequisites
- Having an HrFlow.ai account. To signup, please visit https://hrflow.ai/signup
- Scoring API should be activated
Step 1: Go to the Algorithms Marketplace
The HrFlow.ai Portal is showing two pre-trained Scoring Algorithms:
- Blue & White Collars: general model
- Aerospace Industry: industry specific algorithm
More Scoring Algorithms will be added over time.
Step 2: Choose a pretrained Algorithm
For this tutorial, let's select the Blue & White Collars Algorithm, then click on the button «Create Algorithm».
Step 3: Configure your Algorithm
- Mandatory: Chose a recognizable Algorithm name that you can easily remember, and other users can easily recognize.
- Mandatory: Write a clear Algorithm description that explains the collection of job offers that will be imported in your board.
- Optional: If you want to hide your Algorithm from your teammates, you can make the Algorithm private.
Finally, click on «Create Algorithm» at the bottom left.
After creating your Board, it will be available in Left Sidebar > AI Studio > My Scoring Algorithms.
Scoring API requires the
algorithm_key
To Score profiles or jobs with the API, you must specify an
algorithm_key
in your API request. You will find it in the Overview section of the selected Scoring Algorithm.
Advanced Topics
1. Debugging a Scoring Algorithm (TODO)
2. Auditing a Scoring Algorithm (TODO)
3. Changing the settings of a Scoring Algorithm
Each Algorithm has custom parameters. The 'Settings' section allows you to edit them.
4. Archiving a Scoring Algorithm
You can click the button «Archive this algorithm» on the bottom right.
Archiving a Scoring Algorithm
This operation is irreversible, and all the hyperparameters related to the training of the Scoring Algorithm will be definitively lost.
Updated over 1 year ago