> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/sdgm/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/sdgm/_mcp/server.

# Quickstart: KumoRFM

In this guide, we walk you through an end-to-end example using a multi-table E-Commerce dataset. You'll learn how to load data, auto-infer a relational graph, and run instant predictions: demand forecasting, churn prediction, item recommendations, and missing value imputation, all without training a model.

## Step 1: Install the SDK

Install the Python SDK

```python
pip install kumoai
```

## Step 2: Get an API key

You need an API key to make calls to KumoRFM. Use the API key and API URL provided for your KumoRFM environment.

## Step 3: Initialize a client

Use your API key to initialize a Kumo client:

```python
import kumoai.rfm as rfm, os

os.environ["KUMO_API_KEY"] = "ENTER_YOUR_API_KEY_HERE"

rfm.init()
```

## Step 4: Import your data

We will use an E-Commerce dataset for this example:

```python
import pandas as pd

dataset_url = "s3://kumo-sdk-public/rfm-datasets/online-shopping"

users_df = pd.read_parquet(f"{dataset_url}/users.parquet")
items_df = pd.read_parquet(f"{dataset_url}/items.parquet")
orders_df = pd.read_parquet(f"{dataset_url}/orders.parquet")
```

## Step 5: Create a graph

Auto infer links between tables to form a graph

```python
graph = rfm.Graph.from_data({
    "users": users_df,
    "items": items_df,
    "orders": orders_df,
})

# Inspect the graph - requires graphviz to be installed
graph.visualize()
```

## Step 6: Make a prediction

Write a predictive query to generate instant predictions:

```python
model = rfm.KumoRFM(graph)

# Forecast 30-day product demand
query1 = "PREDICT SUM(orders.price, 0, 30, days) FOR items.item_id=1"
result1 = model.predict(query1)
display(result1)

# Predict customer churn
query2 = "PREDICT COUNT(orders.*, 0, 90, days)=0 FOR users.user_id IN (42, 123)"
result2 = model.predict(query2)
display(result2)

# Item recommendation
query3 = "PREDICT LIST_DISTINCT(orders.item_id, 0, 30, days) RANK TOP 10 FOR users.user_id=123"
result3 = model.predict(query3)
display(result3)

# Missing value imputation
query4 = "PREDICT users.age FOR users.user_id=8"
result4 = model.predict(query4)
display(result4)
```

## Notebooks

Explore practical examples and use cases through our interactive notebooks.

Use KumoRFM for instant predictions using your own dataset with no model training required.

**KumoRFM · pandas**

Build an e-commerce agent that identifies churn risk, predicts probabilities, and generates re-engagement emails.

**KumoRFM · CrewAI · OpenAI · MCP**

Build a sales agent for lead scoring that identifies high-value prospects and creates prioritized outreach lists.

**KumoRFM · OpenAI Agents SDK · MCP**

Build an insurance agent that predicts policy expiration risks, recommends bundling, and generates retention emails.

**KumoRFM · LangGraph · Anthropic · MCP**

Comprehensive guide showing graph schema definition, predictive queries, and model evaluation using Formula 1 data.

**KumoRFM · RelBench · scikit-learn**

Interactive tutorial demonstrating predictive analytics on multi-table datasets using Super Mario Maker game data.

**KumoRFM · pandas**

Learn PQL to frame entire ML tasks in SQL-like statements using the Steam gaming dataset.

**KumoRFM · PQL**

Step-by-step guide for evaluating KumoRFM performance on custom datasets using rel-bench.

**KumoRFM · RelBench · scikit-learn**

Advanced evaluation on complex enterprise ERP dataset for multi-class classification.

**KumoRFM · Hugging Face · datasets**

Demonstrates binary classification on single tabular data using the scikit-learn breast cancer dataset.

**KumoRFM · scikit-learn · NumPy**