How to Use DataRobot

Getting started with DataRobot

In this guide, you will learn how to use DataRobot to build and deploy predictive models without needing extensive data science knowledge. You'll be able to prepare your data, run models, and evaluate their performance quickly.

Step 1: Sign up and set up

1. Go to DataRobot's website.
2. Click on the “Get Started” button located at the top right.
3. Fill in the required fields for your account (name, email, company).
4. Check your email for a verification link and follow it to activate your account.
5. Log in to your DataRobot account.

Step 2: Your first model

1. Once logged in, click on the “New Project” button on your dashboard.
2. Upload your dataset by clicking “Upload Data” and selecting your CSV or Excel file.
3. After your data is uploaded, click “Start Project”.
4. DataRobot will automatically analyse your dataset. Once complete, click on “Start Model”.
5. Choose your target variable (the outcome you want to predict) from the dropdown menu.
6. Click “Next” and select your desired modelling options. Then click “Start”.
7. Wait for DataRobot to build models. Once finished, you will see a leaderboard of models ranked by performance.
8. Click on a model to explore its details and metrics.

Step 3: Get better results

• Use the “Feature Engineering” tab before running models to improve your dataset. This allows you to create new features that might enhance model performance.
• Experiment with different target variables if applicable, as this can yield better results.
• Set up a validation strategy in the “Validation” section to ensure your model generalises well to new data.

Pro tip

Use the “Auto-ML” feature to run multiple models simultaneously. This saves time as you can compare different algorithms and select the best-performing one without manually configuring each model.

Common mistake to avoid

Avoid using datasets with missing values or incorrect data types. DataRobot may not handle these well, leading to suboptimal model performance. Always clean and preprocess your data before uploading.