Build performant predictive models with the TIM engine | Tangent Works

The Tangent Information Modeller

Meet TIM

TIM, or Tangent Information Modeler, is an automatic predictive model building engine that automates the forecasting and anomaly detection process by analyzing time series data and generating accurate models based on the patterns it detects.

KEY FEATURES

Information is valuable, but getting it from data is often difficult and requires expertise. One of the areas where this is particularly true is time series analysis, not in the least because of its highly dynamic nature. Many use cases come down to forecasting and/or anomaly detection with time series, exactly the subjects in which TIM excels. TIM reduces the need for valuable resources (expertise, time and money) and helps users to leverage the insights hidden in their data.
Many modelling tools require tedious, manual “feature engineering” that calls for the work of a domain expert. This results in a compute-intensive trial-and-error process to determine which combinations of input data will be relevant, measured over which time delays. This process can take a long time, especially since the number of combinations grows exponentially with the number of input variables and time intervals. TIM is different, as it automates the feature engineering process described above.
Many models currently in use were built and tuned manually with specialised efforts. TIM’s thorough automation allows the engine to create a model with equivalent or even better accuracy in a matter of minutes.
TIM is built to be compute-efficient and can process huge datasets in minutes time. This means users can automatically create many models and update those models as frequently as desired. By automating the most tedious aspects of model building and fully leveraging existing resources, TIM helps organisations to scale to handle the increasing demand for predictive models.
A great challenge of predictive modelling is to explain the model(s) to the people who depend on it. TIM presents its models as an equation in a human-readable format, which can be visualised as a map that shows the weights of and relationships between the model’s input variables.
TIM’s architecture is optimised for smooth integration with existing databases, BI tools and other enterprise applications. To realise this, all of TIM’s functionalities are easily accessible through its REST API. This structure also guarantees the utmost deployment flexibility.

InstantML

The unique InstantML technology for time series data brings together feature engineering, model building and model deployment. This allows users to obtain high quality results with minimal required effort.

Machine Learning Managing Complexity

1. Handcrafted models

Often, data scientists create handcrafted models to see to companies’ analysing needs. In crafting these models, they go through the processes of data preparation, model building, model deployment and execution. It quickly becomes clear that this is a difficult and laborious task, and while results may be good, they are not scalable.

2. AutoML

In come AutoML techniques, which train many different models and then select the most successful out of them. While this is a great step towards automation, tedious, manual feature engineering is still required, which calls for the input of a domain expert. The result is a compute-intensive trial- and-error process that can take up weeks of valuable time.

3. InstantML

TIM goes beyond these methods with its InstantML technology. With InstantML, a model is created in one single step, including feature engineering, model building and model deployment. All of this is automated and happens under the hood of the TIM engine. The user provides input data and TIM responds with the necessary models, that can be applied to the desired forecasting and anomaly detection purposes.

4. RTInstantML

With the unique RTInstantML technology, TIM goes even further. RTInstantML automates everything from model building to model application, taking input data and directly providing the user with the relevant forecast. This eliminates the need to set up a particular data availability situation, as the situation can be extracted directly from the input dataset at the time of forecasting.

Machine Learning Maturity

Thanks to TIM’s thorough automation of the model building process, no data scientists are required to get great results. Rebuilding and retraining models becomes easy and users can even leave model storage completely out of the picture with TIM’s RTInstantML technology. All of these features are brought together into a scalable architecture.

Deployment

TIM can be deployed in various ways to meet your requirements. You can deploy TIM in the cloud or on-premises or even in the edge on IoT devices.

Tangent Works Cloud

TIM can be used as a cloud service managed by Tangent Works. This offers you ease of use and deployment.

Your Cloud

TIM can be deployed on your own Microsoft Azure or AWS subscription.

Your Infrastructure

TIM engines and applications can be deployed on your own IT infrastructure. The state of the art container-based architecture allows for easy installation and integration in your environment with minimal effort.

IoT Edge

TIM can be deployed on various IoT edge devices. This allows you to bring model building, advanced forecasting and anomaly detection to the field.

Interfaces

After deployment, TIM can be used through various interfaces:

TIM’s API:
For users who desire full control over TIM’s modelling parameters

TIM Studio:
For users looking for a more streamlined graphical interface

TIM Clients or TIM on platforms:
For users who want to implement TIM directly in their favourite tool or platform

TIM Studio

Take a tour – A walk through the intuitive interface of TIM Studio

Slide 1

Zoom in to your data and get a feel of what you are working with.

Slide 2

Iterate over forecast to create the optimal models for your business challenges.

Slide 3

Detect anomalous behavior and dive into the insights behind it.

Slide 4

Organize your analytics work in intuitive structures mapping the underlying business.

Business Roles

TIM for Executives and
Business Managers

The Challenge — How to leverage AI technology to improve operational efficiency.

Forecasting and anomaly detection are essential for most businesses since managers must always try to match supply and demand and make their operations run as smoothly and


efficiently as possible to stay competitive. But most businesses don’t have adequate resources or expertise to leverage AI technology for this purpose, so they often rely on crude manual methods.

For industries like Energy, Utilities and Manufacturing there is added pressure to make sure that the forecasts are both accurate and timely. For example, for one energy supplier a 1% improvement in long-term demand forecasting accuracy translates to a savings of 1GW of peak load capacity and €500K per year.

TIM delivers — A practical way to transform your business
with AI

TIM provides you with timely, accurate forecasts that help avoid the costs associated with system imbalances, reduce system downtime and optimize maintenance cycles. What’s more, thanks to TIM’s automatic predictive model generation you can apply predictive analytics broadly in your organization, and you won’t need to hire and train an army of data scientist to get the job done.

It’s no wonder that most TIM sales leads come from executives who are seeking a practical way to transform their businesses with AI.

TIM for Business Analysts

The Challenge — Meeting the demand for AI-based
forecasting and anomaly detection.

Your organization would like to jump on the AI bandwagon but there is a shortage of data scientists and there’s a tremendous backlog of projects awaiting resources. You’d like to be able to innovate and leverage AI/machine learning technology without needing to be an expert.

TIM delivers — Accessible, accurate AI with fast time to
value.

TIM automates the model generation process so you can create and update predictive models without being an AI expert. So instead of using one generic model to fit a class of customers or equipment, you can create individualized models for each customer or asset, and get the best possible accuracy.

TIM for IT Professionals

The Challenge — Hosting compute-intensive AI tools and integrating them with your existing databases and applications.


Some AI tools are very compute-intensive and require a major investment in on-premise hardware or cloud instances. You’ve likely got plenty of existing databases, data historians, BI tools and enterprise applications that would likely need to interface with a new predictive analytics tool. Firewall restrictions may make cloud-based solutions impractical.

TIM delivers — maximum compute efficiency, rapid integration and flexible deployment options.

TIM was architected for a high degree of compute efficiency based on SIMD Instructions (Single Instruction Multiple Data) Most applications require only a single server with limited amount of memory. For applications with hundreds of input variables and huge files of time-series data, TIM scales automatically and near-linearly to handle big data with fast response times.

It’s also built for rapid integration with your existing databases, BI tools and other enterprise applications since all of its functions are easily accessible through its REST API. You’ll have the utmost deployment flexibility with support for cloud-cloud, on-premise and hybrid configurations. Most integration projects take just a few days to complete.

TIM for Data Scientists

The Challenge — Improving your productivity so you can
focus on the most important tasks.

It’s hard to recruit and retain data scientists these days so your job security is better than ever. That’s the good news. But with a long list of potential AI use cases in your company you may not be able to get the help you need to scale up the use of predictive analytics beyond a small number of projects.

TIM delivers — Scalable, automated, understandable AI

TIM Improves your productivity by eliminating repetitive, trial-and-error modeling tasks and by helping you visualize your model with an intuitive heat map that shows the relative influence of all your input variables. Unlike other modeling tools that require you to choose from

hundreds of pre-defined models, TIM creates an optimized model based on the patterns it detects in your time-series data. You’ll spend less time doing feature engineering, building and explaining models and more time on exploiting the power of predictive analytics across your organization.

Request a Demo

Interested in trying TIM out?

Get in Touch

Let’s explore together how you can get more out of your data.