Embedded Files
One-to-one: Train a new model.
One-to-one: Train a new model.
Training new models to improve the performance of your one-to-one product recommendation engine.
Training new models to improve the performance of your one-to-one product recommendation engine.
1. Introduction
1. Introduction
At the Neuronal networks tab, you can manage the distinct models available to access to the deep learning product recommendation engine.
There are two main actions associated with this tab:
Train a new engine. Is used to train a new recommendation model.
Deploy an engine. Is used to get a production API for a recommendation model.
Your engine information is at the first table shown on the main box.
The information for the engine is: slug (identifier), status, creation date and last update date.
On the Trainings, versions and productive engines list you can see the trainings available for your deep learning intelligence.
Click on [+] on the list to get all the information about one training.
An active training can be used on the bulk one-to-one recommendation.
To use the one-to-one recommendation in real time on your eCommerce or App you need to deploy the training. Deploying the training you are able to use the getRecommendation API call in real time.
Click here for more information: One-to-one used in real time.
Click here for more information: GetRecommendation API call.
2. Train a new engine
2. Train a new engine
Click on the Train engine button on the top-right of the main box.
Complete the form to start the new training:
Write a unique training name.
Select one-to-one engine.
Click on the Train engine button.
Once the training is finished you will be able to deploy this training version and get the production API.
3. Get a production API
3. Get a production API
In order to get a production one-to-one recommendation API you need to deploy your training version engine.
Click on the Deploy engine button on the top-right corner.
Complete the form to deploy the training version engine:
Write the deploy slug. This value must be unique and will be used on the API calls. Click for more information on the getRecommendation API call.
Select the training version you want to deploy. The selected training version must be with the active status.
Click on the Deploy engine button.
When the deployed engine has the active status can be called by the API.
4. Data set
4. Data set
This engine is where the data needs to be ingested in order to make all the calculations for the neuronal network. The main information can be related to:
Products
Customers
Purchase
All of the data is essential to get an accurate recommendation. In order to assure that all the information is correctly updated and evaluated the green SUCCEEDED button needs to be shown.
Page updated
Report abuse