Elevating Time Series AI with Tangent Works and Databricks

We are excited to unveil a development that addresses our customers' top request: the integration of Tangent, our Time Series AI engine, as a Databricks runtime. This is poised to transform forecasting and anomaly detection, setting a new standard for the industry.

Unleashing Unmatched Power in Time Series AI

The fusion of Tangent with Databricks is a big leap forward for businesses aiming to unlock the full potential of their data. This integration offers continuous and real-time model (re)building at an unprecedented scale, delivering precise and explainable forecasts and anomalies detected with ease.

Why Databricks + Tangent Works?

At Tangent Works, we have developed a  ML algorithm to automatically predict future trends and detect anomalies. Our unique capability to dynamically (re)build models on the fly sets us apart, turning what typically takes hours or days of knowledge-intensive data science work into a seamless process of real-time feature engineering and model creation. This innovation delivers results 50 times faster and at just a fifth of the usual cost.

‍

Meanwhile, Databricks has spent a decade building a platform that brings enterprise AI and analytics closer to their data, enabling the deployment of intelligent solutions at scale. The platform features an efficient, hyperscalable toolbox including Delta Lakes, Spark, and robust security standards to minimize the time to value for AI.

‍

The Perfect Synergy for Advanced Analytics

Together, Tangent + Databricks offers the perfect synergy for advanced analytics. By integrating Tangent with Databricks' scalable platform, we bring our solution closer to your data, supporting hyperscalability through parallel job processing while maintaining enterprise-level security. Today, forecasting and anomaly detection across all your business-relevant data has become effortless, real-time, and automatic. The days of back-and-forth between data engineers and scientists without clear results are over. Say goodbye to delays and inefficiencies, experience the future of data processing and predictive analytics today with Tangent and Databricks.

‍

Key Benefits:

Tangent + Databricks in action on a Kaggle M5 dataset:

Are you tired of ML projects never making it to production? Do you wonder why data scientists are doing that same ML process all over again without a clear ROI? Do you also want to focus on deployment instead of R&D? See here how Tangent provides the alternative you need to finally get value from Machine Learning!  

‍

Traditional Workflows: From Data to Prep to Features to Models to Scoring to Deployment

In the fast-paced world of machine learning, traditional methods often involve several cumbersome stages—from data preparation to iterative model development, tracking, deployment, and drift detection. This process, while effective, is time-consuming and keeps forecasting capabilities in the hands of expert data scientists, resulting in limited production use cases and a constant struggle to demonstrate ROI.

Streamlined Workflow: From Data to Deployment

With Tangent, the game changes.

‍

Data: Engineers: Automate feature engineering effortlessly with our robust infrastructure.

ML Engineers: Seamlessly manage CI/CD tools to promote models to production.‍

Tangent: Handles the rest, merging model building and deployment into a single, streamlined step.

‍

First, one simply downloads Tangent from the Databricks Marketplace (read more about Tangent on the Databricks Marketplace in our blog article 'Tangent now available on the Databricks Marketplace') and sets up an ML workspace with notebooks to get started. In this experiment, a subset of the M5 Kaggle competition dataset with Walmart sales data is collected. This dataset includes 122 unique combinations of store and item IDs. For each combination, Tangent will generate a detailed forecast.

‍

By integrating feature engineering and model building in real-time, Tangent eliminates the need for traditional tools and processes. This includes tracking and registering models with MLFlow or detecting drift of static models. With Tangent, models are built and deployed instantly, rendering the concept of model drift obsolete.

‍

Efficiency at its best

Tangent immediately creates 122 models on the fly in just 1.7 minutes, completing feature engineering, model building, and scoring for optimal accuracy. Tangent automatically discovers complex relationships between predictors, utilizing advanced techniques like Fourier transformations and seasonal pattern analysis. This comprehensive approach ensures alignment and data availability, all within minutes.

‍

For each unique combination of store and ID number, Tangent's powerful modeling strategy identifies and leverages intricate patterns, providing immediate value from forecasts. ML Engineers can now integrate Tangent’s cutting-edge solutions into AI pipelines, transforming the way forecasts are utilized and delivering instant results.  

Say Goodbye to Tedious ML Iterations

So, if you're tired of waiting for data scientists to perform additional iterations of building extra features, re-evaluating multiple models, and then tracking and registering the entire process only to start over again, there is an alternative.

‍Say hello to Tangent

Experience the revolution in predictive analytics with Tangent and Databricks here.

‍

‍

The integration of Tangent Works with Databricks marks a significant leap forward in time series AI. This powerful combination streamlines forecasting and anomaly detection, transforming these tasks into real-time, highly efficient processes. By eliminating traditional barriers and reducing costs, businesses can now achieve instant, accurate insights from their data, driving better decision-making and operational efficiency.

‍

To gain further insights on the impact of this integration and the latest developments in data and AI, read our blog article from the DataAISummit 2024 2024, where our CGO, Sam Verdonck, and CTO, Simon Mestdagh share their insights.

‍

Ready to elevate your predictive analytics?

Don't let outdated ML workflows slow you down. Embrace the future with Tangent Works and Databricks. Start now and revolutionize your data strategy!

‍

Article by Simon Mestdagh, Chief Technology Officer at Tangent Works

Connect with Simon on Linkedin!