1. This article details how you can make multiple predictions on the trained model together at a time. After uploading your dataset and training the model, the next step is to check future predictions that you are interested in. Our platform gives the user two options - either running single predictions or uploading a file having multiple rows of data and running a batch prediction on the file. This feature is known as Batch Predict and is available on the Export tab of a prediction report. The Export tab helps users to share predictions

  2. Once you are on the report page, navigate to the Export tab and the first section is Batch Predict. Either drag and drop or upload your data, a csv file (within 25MB size). In case you have a larger dataset, you need to divide it into multiple files and upload them separately

  3. The Batch Predict feature already mentions about the number of feature columns that must be present in your uploaded file. For your ease of understanding, we provide an example file that already has the column headers and the prediction column (in blue) that will be automatically filled in with the predicted values. Please note that if you want, you can have more feature columns that were not used during the training of the model, but you must have all the columns that were used during the training of the model. You can download this Sample file and start filling your rows of data directly into that file

  4. We also have the toggle for automatically filling missing values in your uploaded csv file. In case you have any missing values in the feature columns, the toggle when on automatically fills the value with the mean value of that column. Once everything is uploaded properly the platform performs the predictions and notifies when the predictions are ready. You can then download the predictions file and use the refresh button to upload a new batch predict file

  5. The downloaded predictions file will have two columns added at the end, one for precision (for regression problems) / probability values (for classification problems) and one for predicted values

Did this answer your question?