This quick start guide will walk you through Decanter AI’s streamlined model-building to prediction making process.

1. Upload Data

Under the Builder page, click on the “Upload” button to upload a dataset. 

You can also select a previously uploaded dataset from the dropdown list.

Once you’ve uploaded your dataset or selected an existing dataset, click Next.

2. Prediction Target

Select the prediction target from the dropdown list. Depending on the data type in the target column, 

Decanter AI will automatically determine if this is a classification or regression problem. 

For clustering problems, select None (Clustering) as the prediction target. 

3. Feature Exploration

Decanter AI can automatically detect feature data types (numerical, categorical and date-time). 

If needed, users can also manually edit the data type of each feature.

You can unselect the features you don’t want to include in the training. 

*You need to have at least one feature selected to start training

For numerical features, you can select the type of missing value handling for each feature, 

the options are Auto, Mean, Median, 0 and drop row. 

To learn more about this feature and the difference between each option, check out this article _______

4. Advanced Settings

Users can start the Build Model process with Decanter AI’s default settings and get a decent model.

If you need more info on the details of the Advanced Settings, please refer to this article: ＿＿＿＿＿

5. Build Model

Once your dataset, prediction target, features, and advanced settings are finalized, click on the Build Models button to build models.

Go to the Experiment page to view the training progress.

When the training is done, you will see the training time displayed under Status.

Click on the experiment to reveal the models ranking and the results. 

The model with "Recommended" is more accurate and stable compared to other models.

6. Make Prediction

Users can click on any of the trained models to make predictions.

Select your test dataset and merge ID column from the data set, then click on the Predict button. 

You can see the prediction progress during this process.

Once the prediction is complete, you can click on the View button under Chart to view charts related to this prediction.

You can also download the prediction results by clicking on the More -> Download button under Action.