Predictions
Once a model has been trained, you can use it to generate predictions on new data. DashAI supports two prediction modes: Dataset Predictions and Manual Predictions. Both are accessed from the PREDICTIONS tab inside each model card.
Accessing Predictions
- Open a session from the left sidebar.
- Expand a model card that has a Finalizado status.
- Click the PREDICTIONS tab.
The tab is divided into two sections:
| Section | Description |
|---|---|
| Dataset Predictions | Run the model against an entire dataset loaded in DashAI. |
| Manual Predictions | Enter one or more rows of data manually and get an immediate prediction. |
Each section shows a count badge with the number of predictions already created, and a button to add a new one.
Dataset Predictions
Use this mode when you want to apply the model to a full dataset — for example, to generate predictions for a test set or a new batch of records.
Creating a Dataset Prediction
Click NEW DATASET PREDICTION to open the Create New Prediction modal. The modal has two steps.
Step 1 — Configure Input
- Select a dataset — Choose from the datasets available in DashAI. The dropdown
shows the dataset name and row count (e.g.,
Dataset_1 (45000 Rows)). - Once a dataset is selected, a Prediction Information panel shows:
- Input Columns — the feature columns the model expects, shown as tags.
- Target Column — the column the model will predict, highlighted in teal.
- A preview table shows the first rows of the selected dataset so you can verify it has the right structure before proceeding.
Click NEXT to continue.
Step 2 — Confirm
A summary card shows the prediction configuration:
- Model — the algorithm type.
- Run — the specific model instance name.
- Input Type —
Dataset. - Dataset — the dataset selected in Step 1.
Review the summary and click SEND to queue the prediction job. Click BACK to return to Step 1, or CANCEL to discard.
Prediction Result
Once complete, the prediction appears as a card under Dataset Predictions with:
- Prediction #N — sequential number.
- Timestamp — date and time the prediction was run.
- Dataset name — the source dataset used.
- Finalizado badge when processing is complete.
- A preview table showing the first 100 rows of results.
- A note: "Preview of first 100 rows. Download the CSV for complete results."
- A download button (↓) to export the full results as a CSV file.
- A delete button (🗑) to remove the prediction.
The preview shows 100 rows — always download the full CSV when working with large datasets to access all prediction results.
Manual Predictions
Use this mode when you want to test the model on specific, hand-crafted input values — for example, to understand how the model responds to a particular combination of features, or to demonstrate the model's behavior in a classroom or presentation context.
Creating a Manual Prediction
Click NEW MANUAL PREDICTION to open the Create New Prediction modal.
Step 1 — Configure Input (Manual Data Entry)
The modal shows a table with one row per prediction instance. Each column corresponds to an input feature the model expects.
- Categorical columns appear as dropdowns with the valid category values.
- Numerical columns appear as editable text fields.
Fill in the values for each column in the row.
Click Add Row (the + button) to add additional rows if you want to predict
multiple instances at once.
Click NEXT to continue.
Step 2 — Confirm
A summary card shows:
- Model — the algorithm type.
- Run — the specific model instance name.
- Input Type —
Manual Input. - Manual Rows — the number of rows you entered.
Click SEND to submit. Click BACK to return and edit the data.
Prediction Result
The result appears under Manual Predictions with the same structure as dataset predictions: a numbered card with a timestamp, Finalizado badge, and a preview table showing the input values alongside the predicted output.
A download button (↓) and delete button (🗑) are available on each result card.
Tips
- For Dataset Predictions, the selected dataset must have the same input columns (same names and types) as the dataset the model was trained on. If column names differ, the prediction will fail.
- Use Manual Predictions to quickly validate that the model's behavior makes intuitive sense on specific examples before running it on a full dataset.
- You can create multiple predictions from the same trained model — each is stored independently and can be downloaded separately.
- Predictions persist in the session even after you close and reopen DashAI.
Troubleshooting
| Symptom | Likely cause | Solution |
|---|---|---|
| PREDICTIONS tab is empty or disabled | Model has not been trained | Train the model first — the tab activates after a successful run |
| Dataset prediction fails | Column mismatch between training and prediction dataset | Ensure the prediction dataset has the same input columns as the training dataset |
| Manual prediction form is missing columns | Model configuration has changed since training | Re-open the model and verify the input columns in the session setup |
| Download button produces an empty file | Prediction job did not complete successfully | Check the Job Queue for errors and re-run the prediction |