Sign in Start for free
sub header image
Support Library
Web Prediction
This functionality is not implemented for this data type.
Web Prediction (for disabled DSP only)
To predict using new data you can use a web interface. The service can be enabled by clicking “Enable”. Web prediction requires cloud infrastructure to process new data, deploy models and calculate predictions. Enabling the prediction service will result in some infrastructure charges.
Enable Prediction
At the time these infrastructure charges are accepted, monthly bills will be generated and delivered in accordance with our pricing policy.
After the service is enabled, the “Web Prediction” and “REST API Access” buttons will become active:
Start Button
Click “Start” next to “Web Prediction” to upload new data.
Upload New Data
The dataset uploading/selection process is similar to the process described in the “1. Select Data for Training” section. The test dataset (new data) must have the same structure as the training dataset and meet other dataset requirements. Even if there is a target variable in the dataset, the platform will ignore it. For preloaded training datasets, the test dataset is preselected.
After the dataset is uploaded/selected, click “Start” to start the prediction process. When the prediction is completed, the system will display the following message:
Prediction Completed
Users may choose to “Download” the predictions to a local hard drive as a CSV file or “View” the prediction results in the browser window. If “View” is selected, the predictions will appear concatenated to the existing data (uploaded for prediction):
Preview Prediction
You can see the predicted values in the first column, as well as additional information:
For binary and multiclass classification task types – the probabilities of the predicted classes.
For the regression task type – the Confidence Interval and Confidence Probability.
For any task type Neuton calculates the Model-to-Data Relevance Indicator for:
every row sent for predictions
dataset sent for predictions
all data sent for predictions aggregated over time (Historical Model-to-Data Relevance Indicator). Displayed on the Prediction page in the information area.
The structure of prediction results is the same for any kind of prediction (WEB, REST API, or using Downloadable Solution).
If you previously removed variables in the training dataset using the Remove variables option but did not exclude them from the test dataset, they will also be presented in the results, with a note that they did not participate in predictions.
At this step, a Model Interpreter is also available. To do this, you need to hover over the line you are interested in and click on the “Model Interpreter” button that appears. The “exit” icon next to a row in the list of all predictions indicates that you have already opened the Model Interpreter for that row.
Model Interpreter
All previously made predictions are displayed in a list under the Start button.
List of Predictions
Web Prediction
To predict using new data you can use a web interface. The service can be enabled by clicking “Enable”. Web prediction requires cloud infrastructure to process new data, deploy models and calculate predictions. Enabling the prediction service will result in some infrastructure charges.
Enable Prediction
At the time these infrastructure charges are accepted, monthly bills will be generated and delivered in accordance with our pricing policy.
After the service is enabled, the “Web Prediction” and “REST API Access” buttons will become active:
Start Button
Click “Start” next to “Web Prediction” to upload new data.
Upload New Data
The dataset uploading/selection process is similar to the process described in the “1. Select Data for Training” section. The test dataset (new data) must have the same structure as the training dataset and meet other dataset requirements. Even if there is a target variable in the dataset, the platform will ignore it. For preloaded training datasets, the test dataset is preselected.
After the dataset is uploaded/selected, click “Start” to start the prediction process. When the prediction is completed, the system will display the following message:
Prediction Completed
Users may choose to “Download” the predictions to a local hard drive as a CSV file or “View” the prediction results in the browser window. If “View” is selected, the predictions will appear concatenated to the existing data (uploaded for prediction):
Preview Prediction
You can see the predicted values in the first column, as well as additional information:
For binary and multiclass classification task types – the probabilities of the predicted classes.
For the regression task type – the Confidence Interval and Confidence Probability.
For any task type Neuton calculates the Model-to-Data Relevance Indicator for:
every row sent for predictions
dataset sent for predictions
all data sent for predictions aggregated over time (Historical Model-to-Data Relevance Indicator). Displayed on the Prediction page in the information area.
The structure of prediction results is the same for any kind of prediction (WEB, REST API, or using Downloadable Solution).
If you previously removed variables in the training dataset using the Remove variables option but did not exclude them from the test dataset, they will also be presented in the results, with a note that they did not participate in predictions.
At this step, a Model Interpreter is also available. To do this, you need to hover over the line you are interested in and click on the “Model Interpreter” button that appears. The “exit” icon next to a row in the list of all predictions indicates that you have already opened the Model Interpreter for that row.
Model Interpreter
All previously made predictions are displayed in a list under the Start button.
List of Predictions


Stay updated, join the community