I started my career as a management consultant. Excel was our temple for analytics. For a recent graduate, macros and connected models perform miracles but albeit at great effort. However, our reach was extremely limited compared to the possibilities of today. We could not process anything with images, text, audio or video easily as non-technical users. Fast forward to today, citizen data scientists are unleashing machine learning on all of companies data to run diagnostic, predictive and prescriptive analytics.
How is AI contributing to analytics capabilities?
Thanks to the latest advances in AI, analytics is becoming
- more efficient thanks to automation
- more accessible thanks to improved UI. Natural Language Processing enables analytics tools to understand natural language queries
- more powerful since previously difficult to analyze data such as text and videos are now easily analyzable
Analytics is getting automated
Analytics is time consuming and analytics talent is expensive. The war for AI talent is well documented and impacts the cost and availability of analytics talent. Data scientist who compromise a significant share of modern analytics work force are also part of the AI workforce.
Therefore automating analytics tasks has significant potential value for firms.
Analysis (i.e. discovering insights and acting on them) is getting automated
AI systems are able to analyze data autonomously. Based on the results of analysis, they can take automated actions or highlight insights to employees who can decide the best course of action.
Setting up auto notifications based on simple rules has been possible since the early days of computing. Now companies can set up complex, machine learning based triggers to identify insights or automate actions. For example, a machine learning system that detects intruders to physical secure locations based on video feeds can take actions (e.g. by informing the intruder) or highlight the event to the human personnel.
Report preparation is getting automated, making analytics more accessible
Analytics are only as useful as long as it is accessible. Natural language generation (NLG) enables automated report preparation.
Analytics is becoming more accessible
We mentioned the increasing cost of analytics talent due to the increased demand for data science talent. Data scientists are expensive as they are PhDs or graduates of computer science and other quantitative fields who have an understanding of statistics and computation. This enables them to build complex models including machine learning models which can extract insights from data.
What if you did not need data scientists to extract insights from data? Users can use natural language to easily and intuitively find answers. This is supported through natural language (NL) query (also called natural language interaction – NLI or natural language user interaction – NLUI). For example, Thoughtspot, which enables companies to run complex analytics queries via natural language, raised ~250m on a ~2 billion valuation in 2019.
Such advances enable democratization of analytics and citizen data scientists to tackle large volumes of data fast.
Scope of analytics is increasing thanks to AI
Unstructured data and personally identifiable information limited the scope of analytics before the advance of AI algorithms but now companies are able to directly or indirectly use these data in their analytics efforts.
Unstructured data is becoming analyzable thanks to AI
Excel and other legacy analytics tools are not effective at dealing with unstructured data such as text, audio and images. Advances in AI greatly expand the scope of analytics when compared to the days when excel was the primary analytics tool. Some ways that AI is becoming integrated in analytics includes these areas:
- Natural language processing (NLP) enables analysis of text
- Transcription enables speech analytics
- Computer vision enables image and video analytics
Semi-structured data is becoming analyzable thanks to AI
A significant share of company data is locked up in semi-structured documents such as invoices, receipts, order forms etc. Deep learning based data extraction solutions enable companies to extract entities from their semi-structured data and use them to understand their business in more detail.
New techniques enable analysis of anonymized personally identifiable data, expanding scope of analytics
Anonymization via synthetic data is a rather old technology. However with the increasing demand for analytics and increasing protection of personal data, demand for anonymized data has increased. Numerous synthetic data vendors are enabling companies to create synthetic (machine-generated, anonymized but following the same distributions as the underlying personally identifiable data) copies of their customers so they can run detailed simulations and improve their offering.
Analysis is becoming more powerful thanks to AI
While simple regressions guided business decision making for hundreds of years, businesses now rely on machine learning. Machine learning is the use of statistical techniques to enable computers to identify and learn the patterns in the given data, rather than being programmed explicitly for a certain function.
Some analytics techniques that can be enhanced with AI and machine learning include:
Prediction: Using short and long term variability in data to enhance forecasting efforts.
Pattern recognition: Understanding normal trends in order to spot anomalies, as is often the case in fraud detection.
Classification algorithms: Grouping and organizing of data, includes clustering.
Which industries rely on AI in analytics?
All industries can benefit from AI-powered analytics. We had a deep-dive into manufacturing analytics highlighting the impact of AI.
What are some analytics tools with integrated AI functionality?
Some tools that can make the analytics process easier include:
|Name||Status||Number of Employees|
|Alphine Data Chorus||Private||11-20|
|Einstein Analytics by Salesforce||Public||10,000+|
|Microsoft Power BI||Public||10,000+|
|Oracle Analytics Clous||Public||10,000+|
|SAP Analytics Cloud||Public||10,000+|
|TIBCO Software Inc||Private||1,001-5,000|