Kumo Engine Examples | NVIDIA Kumo

Browse the notebooks and Python examples in the Kumo Engine examples repository. Choose an example below and follow its setup and data prerequisites.

Examples are grouped by industry, with deployment and scope listed for each workflow. Entries marked Commented preview contain demonstrations that are not runnable with the supported package API.

Cross-industry

Notebooks

Example	Deployment	Scope
Build a custom training table	Local	Focused workflow
Choose a static or temporal query	Local	Focused workflow
Choose the customers to score	Databricks	Focused workflow
Preview graph visualization	Local	Commented preview

SDK code examples

Example	Deployment	Scope
Compare a retrained challenger with a trusted champion on identical held-out rows	Local	Model workflow
Compare neighbor sampling choices on the same CustomerLTV training rows	Local	Model workflow
Configure delimited strings and native arrays as categorical collections	Local	Focused workflow
Fit a users repeat-purchase model and retain its exact bundle for a score job	Local	Model workflow
Fit and retain a CustomerLTV model for retraining in a fresh Python process	Local	Model workflow
Inspect a Snowflake schema with a configured local connector identity	Snowflake	Focused workflow
Inspect rooted and any-URI Amazon S3 connectors using ambient AWS credentials	Local	Focused workflow
Preview interactive inspection of a small synthetic shop graph	Local	Commented preview
Read synthetic CSV and Parquet tables directly and snapshot their graph	Local	Focused workflow
Retrain a retained CustomerLTV model in a fresh Python process using current source data	Local	Model workflow
Schedule daily local Engine batch predictions on an Airflow worker	Local	Focused workflow
Scheduling preview for the MLOps fit/score integration	Local	Commented preview
Scheduling preview for the MLOps fit/score integration	Snowflake	Commented preview
Score current source data with the exact trusted bundle produced by the fit job	Local	Model workflow
Turn prediction explanation prompts into natural-language explanations	Local	Focused workflow

Energy

SDK code examples

Example	Deployment	Scope
Regress hourly transformer oil temperature from the public ETT series	Local	Model workflow

Financial services

Notebooks

Example	Deployment	Scope
Predict reported transaction fraud on Databricks	Databricks	Model workflow
Screen accounts for money muling	Databricks	Model workflow
Screen synthetic transactions for review	Local	Synthetic binary classification
Detect money mule activity in synthetic accounts	Local	Synthetic account classification with a baseline and new-account scoring

SDK code examples

Example	Deployment	Scope
Classify loan status from the eight-table public Financial relational dataset	Local	Model workflow
Predict Home Credit default risk from seven public competition tables	Local	Model workflow

Healthcare

Notebooks

Example	Deployment	Scope
Predict appointment attendance in the next 30 days	Local	Synthetic temporal binary classification

Hospitality

SDK code examples

Example	Deployment	Scope
Classify the booking destination of Airbnb new users using sessions and demographics	Local	Model workflow
Forecast three daily interaction counts for each user	Local	Model workflow
Predict each user’s interaction count for the next day	Local	Model workflow
Predict whether each user will interact in the next day	Local	Model workflow
Predict which interaction types each user will have in the next day	Local	Model workflow
Rank restaurant relationships without a future prediction window	Local	Model workflow
Rank restaurants each user may interact with in the next day	Local	Model workflow
Recommend available recipes from authorized recipe/order tables	Local	Model workflow
Reload a trusted restaurant model and score its unchanged original snapshot	Local	Model workflow
Train a small restaurant model and score twenty held-out cuisine labels	Local	Model workflow

Media and entertainment

Notebooks

Example	Deployment	Scope
Predict low ratings for movies on Snowflake	Snowflake	Model workflow
Predict the content formats each subscriber will use	Local	Synthetic multilabel classification

SDK code examples

Example	Deployment	Scope
Predict a movie’s average rating during the next twelve months	Local	Model workflow
Recommend Stack posts a user will comment on over the next two years	Local	Model workflow
Regress MovieLens 1M ratings for user/movie pairs at their rating times	Local	Model workflow

Retail

Notebooks

Example	Deployment	Scope
Classify store products into a category	Local	Synthetic multiclass classification
Explore Kumo Engine tasks with TPC-H on Snowflake	Snowflake	Model workflow
Kumo Engine task examples with AdventureWorks in Google Colab	Colab	Model workflow
Kumo Engine task examples with TPC-H on Databricks	Databricks	Model workflow
Predict repeat orders with AdventureWorks	Snowflake	Model workflow
Predict repeat orders with AdventureWorks on Databricks	Databricks	Model workflow
Recommend products from synthetic purchase histories	Local	Synthetic temporal recommendations
Tag store products with every applicable label	Local	Synthetic multilabel classification

SDK code examples

Example	Deployment	Scope
Combine two-part spend predictions with explicit row alignment	Local	Focused workflow
Fit H&M lifetime value or recommendation models with supplied example weights	Local	Model workflow
Forecast 28 daily sales values from the public M5 competition	Local	Model workflow
Forecast three 30-day order counts for each customer	Local	Model workflow
Match Walmart and Amazon products using the public labeled entity-matching pairs	Local	Model workflow
Predict 28-day customer churn by sales channel in Snowflake	Snowflake	Model workflow
Predict H&M churn, conditional lifetime value, recommendations or daily revenue	Local	Model workflow
Predict H&M repeat purchase over 90 days and report complementary churn risk	Local	Model workflow
Predict conditional 30-day maximum purchase quantity using the public CustomerLTV mini dataset	Local	Model workflow
Predict each active customer’s order count for the next 90 days	Local	Model workflow
Predict the size of the next Instacart basket within 30 days	Local	Model workflow
Predict whether an active customer orders again in the next 90 days	Local	Model workflow
Predict whether each active customer orders online, offline, or both	Local	Model workflow
Preprocess customer-location entities with read-only Snowflake SQL	Snowflake	Focused workflow
Rank product relationships without a future prediction window	Local	Model workflow
Rank products each customer may purchase in the next 90 days	Local	Model workflow
Recommend coupons from the public Coupon Purchase Prediction competition	Local	Model workflow
Recommend future customers for each product in the public Olist dataset	Local	Model workflow
Recommend stores for each user/location combination after explicit preprocessing	Local	Model workflow
Recommend twelve related H&M articles from a prepared article-pair table	Local	Model workflow
Reload a trusted AdventureWorks model and score a compatible snapshot	Local	Model workflow
Run paired H&M model scripts as local processes and retry only failed fits	Local	Focused workflow
Train an AdventureWorks repeat-order model and score active customers	Local	Model workflow
Train one half of a two-part H&M expected-spend model	Local	Model workflow

Science

SDK code examples

Example	Deployment	Scope
Forecast monthly sunspot counts from the public SILSO series	Local	Model workflow