Step-by-Step Tutorial for Training Your First Model

This guide will walk you through the process of training your first predictive model and making predictions using Predict Oracle. We'll use a sample dataset to demonstrate the entire workflow.

Before You Begin

Make sure you have: - A Predict Oracle account - Access to the dashboard - Sample data ready in CSV format (or use our provided example dataset)

Step 1: Choose Your Model Type

1. From the dashboard, click the "New Project" button
2. Select "Outcome Prediction" as your model type
3. This is the simplest model type for beginners
4. It predicts yes/no or categorical outcomes

Step 2: Upload Your Data

1. Click "Upload Dataset" and select your CSV file
2. For this tutorial, you can download our Sample Customer Conversion Dataset
3. Wait for the upload to complete
4. Verify that your data preview looks correct
5. Column headers should be visible
6. Data types should be properly detected

Step 3: Configure Your Model

1. Select your Target Column
2. This is what you want to predict (e.g., "converted")
3. Choose your Feature Columns
4. These are the attributes used to make predictions
5. For the sample dataset, select: age, income, previous_purchase, email_opens
6. Select your ID Column
7. This uniquely identifies each row (e.g., "customer_id")
8. Advanced Settings (optional)
9. Leave at default values for your first model

Step 4: Train Your Model

1. Click the "Train Model" button
2. Add a descriptive name for your model (e.g., "Customer Conversion Predictor v1")
3. Wait for training to complete (typically 2-5 minutes)
4. You'll see a progress indicator
5. You can close the browser and return later—we'll email you when it's done

Step 5: Review Model Performance

1. Once training is complete, review the performance metrics:
2. Accuracy: Overall correctness of predictions
3. Precision: How often "yes" predictions are correct
4. Recall: Percentage of actual "yes" outcomes correctly predicted
5. Feature Importance: Which factors most influenced predictions
6. Examine the visualizations to understand your model better

Step 6: Make Predictions

1. Click "Make Predictions" button
2. Choose your prediction method:
3. Upload new data: For batch predictions on new customers
4. Interactive prediction: Test individual scenarios
5. For interactive prediction:
6. Enter values for each feature
7. Click "Predict" to see the result
8. For batch prediction:
9. Upload a CSV with the same structure as training data (minus the target column)
10. Click "Submit" and wait for results
11. Download the completed predictions when ready

Step 7: Interpret Your Results

For the sample dataset, your results might show: - Customers with high email_opens are more likely to convert - Previous purchasers have higher conversion rates - Age and income have varying impacts depending on other factors

Next Steps

Congratulations! You've successfully trained your first predictive model. To improve your results: 1. Review our "Data Cleaning Best Practices" guide 2. Experiment with different feature combinations 3. Try advanced configuration options to optimize performance 4. Schedule regular retraining to keep your model up-to-date

For more detailed guidance, check out our model-specific tutorials in the Learning Resources section.