How do I start using MultiViz?
Getting started is easy. Our customer success team will first meet you to make sure the data infrastructure is in place to retrieve the time-series data. The installation can be done remotely. We then offer you a product training that usually takes 2-3 hours.
How to get a profitable “AI solution”?
In analytics, success is dependent upon asking the right questions related to the business, the stakeholders and the KPIs. Framing a business question properly is critical because if you don’t, you could go astray in your analysis or return with the wrong insights.
First questions are what pain points and target KPI’s can improve your business? Then we recommend doing a data quality check to find out if your data contains enough information related to the business case and KPIs you are targeting. MultiViz outputs a data readiness report and our customer success team will help you in this process.
Is it hard to become proficient at using MultiViz?
A subject matter expert can learn most functions of MultiViz within a few hours. The same expert can experiment with MultiViz and become proficient at analysing and annotating data quickly. As for customized data-import, data-preparation and cleansing functions and templates, we offer 50 hours of free customization. And we will continue customization at special rates if you do not have the expertise to write data preparation and cleaning functions.
Is there a minimum requirement in terms of historical data?
We recommend performing a data quality check to find out if your data contains enough information related to the target KPI’s. MultiViz outputs a data readiness report. Our success team are ready to help you if needed.
Do we need to have sensors installed in order to start? Do you offer sensors?
Sensors are not our core business. However, if there is a customer need for more or new sensors, we have reliable partners for both IoT sensors and more demanding industrial applications.
How long does it take to install MultiViz?
The MultiViz installation file is usually downloaded and installed on most business networks and systems within 10 minutes.
What operating systems can MultiViz run on?
MultiViz runs on Windows but can also be hosted in Docker and Linux if needed.
How many fault events are needed in the training data to get insights?
There are different analytics engines implemented in MultiViz. First is an unsupervised event detection algorithm that can be used to highlight and summarize all major changes in the multivariate data sets. This requires no training data. This helps to discover events faster and perform their root cause analysis efficiently.
However, in order to understand the system better and to build more meaningful and reliable applications like anomaly detection or predictive analysis with low false alarms, annotated data or training data is recommended. For instance, for anomaly detection algorithms to perform at low false alarm rates, we recommend having a big pool of annotated data related to the normal behaviour, which is majorly exhibited by the system.
In case, where different modes or events need to be detected and predicted, the quality of detection and prediction will increase with the amount of data related to the events or modes or periods of interest. This can be easily achieved using the powerful annotation and mode identification features in MultiViz.
Do I need to upload my data to the cloud?
MultiViz can be run in the cloud or on our customer premises. Many of our customers want the security of handling their industrial data on their own servers. We support that.
Benefits using MultiViz
Why MultiViz, the self-service advanced analytics software?
In industrial applications the importance of having informative data is crucial for implementing predictive models as well as other data driven solutions. To convert raw sensor data to information requires a subject matter expert (SME).
Our focus is to enable the experts to perform this difficult work much faster and with more accuracy. We enable you to build a holistic context around the diﬀerent variations and patterns in your data, draw faster insights from your data and make it clear!
What is MultiViz?
At the core, MultiViz is a self-service analytics software for the heavy industrial companies that puts the power of data into the hands of those who understand what it means: the subject matter experts.
MultiViz enables SMEs like condition-monitoring, reliability and process experts to rapidly visualize sensor data from assets and processes. It enables SMEs to prepare, analyse and process sensor data without being a data scientist. Using MultiViz, SMEs can operationalize their insights as smart alarms and applications.
How can MultiViz improve production optimization, uptime and overall plant profitability?
MultiViz is a self-service analytics tool that enables to efficiently capture your expert knowledge in the data in order to perform diagnostic analysis and to deploy data-driven machine learning models in your production. To enable this, we have powerful features like event discovery, sensor ranking algorithm for root cause analysis, data-preparation and cleaning templates, operational and failure mode identification, fingerprinting, annotation, knowledge creation and sharing and much more.
Based on your identified business case, MultiViz also helps to build, implement, test and deploy data-driven models for prediction, optimization and automation much more efficiently. You can build your proof-of-concept solution with the smart data, and finally realize industrial-scale solution in a matter of months.
How do MultiViz work with predictions?
Predicting an event means being able to estimate or determine in advance when the event will happen, with a certain confidence. Most prediction algorithms with high performance and reliability require high quality training data.
MultiViz is a powerful tool for enabling SMEs to prepare high quality training data or annotated data. This simplifies building reliable and scalable predictive applications efficiently.
What are MultiViz main features?
MultiViz has a broad spectrum of features that makes it powerful for analysing complex production data.
- Connect: MultiViz connects to all industrial databases. Sensor tags of interest can be immediately retrieved from all databases. All the data can be merged, synchronized, and visualized.
- Clean: With built-in templates for data preparation and cleansing, experts can head start their analysis directly.
- Discover: Discover what is important in the data. Discover interesting events and abnormal patterns which are especially efficient for multidimensional systems.
- Diagnose: Understand what happened and why they happened with our unique sensor ranking algorithm.
- Fingerprint: Create data-driven signatures for relevant events and search for similar events in the data.
- Mode identification: Automatically identify different operational modes and failure modes.
- Predict: Predict asset and process health based on the modes identified, events discovered and other expert annotations
- Contextualize: Build a holistic context around the diﬀerent events discovered. Combine the events’ summaries, sensor ranking lists, expert annotations and the sensor data plots in order to understand the data and contexts much better and faster.
- Capture: Capture core expert and domain knowledge with a health history of the assets and processes. Save it in case logs connected to each asset or use case.
- Collaborate: Share experts’ insights with the other experts and stakeholders in the organization for joint-analysis and troubleshooting. Collaborate with data-scientists to build ML applications with expert knowledge. All knowledge is stored in one place.
- Operationalize: Operationalize experts’ insights as real-time ML applications including smart alarms, anomaly detection, predictive maintenance, optimization.
- Scale-up: With MultiViz, go from a machine learning based PoC to a scaled-up industrial AI solution in a matter of months.
What kind of patterns can be discovered or recognized?
MultiViz can be used for analysing time-series data, spectral data and spatial data. It can be used to understand how fast the multivariate data varies and why it varies over time. It can be used to discover specific patterns and variations of interest and much more.
What is Mode identification?
Most industrial machines or processes have multiple operational and failure modes. Different modes have different characteristics that can be identified, learnt and applied to new dataset that have not analysed. This is one of the core features of MultiViz, which supports automatic detection and recommendations of multiple modes.
Can I specify normal and abnormal periods of data?
MultiViz can be used to identify the multiple normal periods, multiple failure modes and abnormal operating modes with minimal inputs from the SME. The so-identified modes and periods can be learnt and applied to new dataset with little a priori expert knowledge.
How can I adapt MultiViz to my own special needs?
Customization is at the core of MultiViz. Be it data connectors to your industrial databases, data preparation and cleaning methods, data analysis algorithms, almost everything can be customized and reused in MultiViz’s flexible plug-in architecture. The plug-in architecture allows custom scripts in R or Python or other scripting languages to be incorporated with ease.
How do I save cost and time by using MultiViz?
You can save money and time in four steps using MultiViz.
- Our customer cases show that the annotation can be completed 10 times faster when the SME’s add their expert knowledge to the data using MultiViz. You save 4-8 weeks’ work time in some of the common industrial use-case that our customers have worked with.
- Faster and efficient root-cause analysis and event discovery can help reduce several weeks of work to a few hours.
- Save several hundreds of hours in building and deploying customized data driven application compared to a generic platform or having consultants design it for you. Realize scaled up AI solutions in a matter of months.
- Before building and deploying a smart application (optimization, automation or predictive maintenance solution) there is always a good business case that increases your revenue. The early stage data readiness report makes sure that no useless solutions are designed. With MultiViz, you can adopt a data-driven approach for business use case identification.
Start-up and run MultiViz
How does MultiViz help me understand the data?
Industrial data often consists of tens and even hundreds of data channels. It can be very time consuming and sometimes impossible for the human eye to understand the cause of an event. MultiViz automatically finds and zooms into the events of interest and helps to figure out the most important sensors contributing to the event. Root cause analysis has never been faster.
Furthermore, our mode identification algorithms help to automatically identify and annotate operational modes, failure modes and other abnormalities. When you and your colleagues label or annotate the events, everyone in the organisation will benefit from easy understandable data.
How can we benefit from MultiViz as a group/team?
MultiViz is designed to be a company-wide tool to share and democratize your asset data as well as your expert knowledge. The tool consists of a project dashboard where different engineers and experts can interact in parallel. All data and insights can easily be shared between colleagues.
How can MultiViz improve collaboration in my organization?
Collaboration is a key feature in MultiViz. Our software enables your experts to share insights with all stakeholders in the organization for joint-analysis and troubleshooting. You can collaborate with data-scientists to build machine learning applications with expert knowledge. MultiViz supports efficient report generation in PowerPoint and other formats.
How quickly can I get value from my data after I start using the MultiViz system?
We help you with getting started in a few hours, and depending on the use case, you can get insights in as little as minutes or hours. Our experts will help you with any needed database connectors, data preparation and cleaning functions and guide you through the software features. We have seen that users often start to capture insights and add their own knowledge directly.
Can MultiViz prevent knowledge loss at retirements?
MultiViz is a user-centric software where projects and assets can be organized easily. The tool works as a common platform for the company where all colleagues can contribute with their expert knowledge and share information and insights. It includes efficient tools for adding subject matter expert knowledge to your asset data and store that for future generations.
About Viking Analytics
Who are Viking Analytics?
Founded in 2017, we at Viking Analytics publish software for manufacturing, chemical and process industries to harness the power of their process manufacturing data and enable predictive operations. By partnering with us, industrial companies accelerate their digital transformation. Headquartered in Gothenburg Sweden, we are a venture-backed privately held company. At the core of our technology offering, we strive to address the following questions.
- How can companies and their experts find opportunities for optimization and cost saving in their data?
- How do organizations and their experts understand the quality of their data and the information in it?
- How can their experts convert raw sensor data to smart-sensor data with their valuable domain-knowledge in a scalable way?
- How can companies transition from a PoC ML application with smart data to a scaled-up industrial AI solution?
What products and services do you provide?
Our core product is a self-service analytics software, MultiViz, which enables the subject-matter experts and specialists to prepare, analyze and process large sensor data without being a data scientist. The subject matter experts include process engineers, reliability engineers and condition monitoring experts, and we put the power of data into the hands of these people who understand what it means.
With our solution, experts are enabled to convert raw sensor data into smart data and harness the power of process manufacturing data without needing advanced data engineering skills.
Further building on our core is our unique, customizable software for real-time deployment and testing called MultiViz Integrator. This is a data-science-team-in-a-box software solution, which helps organizations to build, deploy and manage data science algorithms and AI models. This help organizations to build scalable and reliable smart manufacturing solutions including optimization, predictive maintenance, anomaly detection, digital AI operators etc. This helps save several months for solution deployment, testing, logging and version management.
With our Data-science-team-in-a-box solution, we enable traditional organizations to realize market-ready data-driven software solutions in a matter of months.
What is unique about your offer versus other companies?
Viking Analytics enable organizations and their experts to realize the full potential hidden in their production data. Our tools, solutions and services enable industrial companies to convert raw sensor data to smart sensor data with expert knowledge efficiently. Further, our tools enable to build proof-of-concept solution efficiently with the smart data, and finally realize the industrial-scale solution in a matter of months. More specifically, our strengths and uniqueness are the following:
- We enable capture of expert knowledge in sensor data with our software for non-data-science experts.
- We enable seamless collaboration between data engineers & SMEs to build ML models.
- We enable customers to scale PoC AI solutions to full-scale industrial system solutions in a matter of months