Introduction to AI Insights in Power BI Desktop

[PriscillaCamp]

Introduction

I am sure you have wondered what AI Insights is; and I am also sure you have wondered (especially if you are not familiar with AI) what the vision, text analytics, and Azure ML functions are. If you are not familiar with it, you can find this group in the Home and Add Column tabs, as shown below

It is my plan to discuss with you today about these AI Insights in Power Query. Because there is so much to talk about, I have decided to break my blog into four parts to make it easier to digest all the content. In this blog, I will talk about what AI is and then give you some background information on some fundamentals to help you understand the AI functions. The remaining three parts will each be discussed separately: Text Analytics, Vision, and then Azure Machine Learning.

Artificial Intelligence and its key elements (sub-fields)

To start, I believe many of you are already familiar with AI by seeing it in sci-fi movies or even on the Internet. Although you probably have an idea from this, you probably may not be familiar with the technical definition or its actual uses. Therefore, I feel it would be good to start with a definition that is very clear and easy to understand. One that fits this requirement to me is Microsoft Learn’s definition of AI, “the creation of software that imitates human behavior and capabilities.” We can see that Artificial Intelligence attempts to mimic what we do, but don’t worry computers won’t be replacing us anytime soon 😉

Now let’s look at the five key elements that are considered a part of AI. It’s important to know these five elements because they will help you understand what the specific features of AI can do. It will also help you do further research if your project needs further analysis. I have listed 5 below and bolded the three that are part of the AI Insights group.

1. Machine Learning – Uses algorithms to teach a computer model to make better decisions from data. The model will improve as it gains experience and learns from its mistakes. There are three types of machine learning: supervised, unsupervised, and reinforcement learning. An example of using machine learning is to determine if a patient is diabetic or not.

2. Anomaly Detection – is used to identify atypical patterns in data. You can think of credit card fraud – when there is an unusual transaction, you will receive an alert from your bank.

3. Natural Language Processing – the computer has the ability to understand written and spoken language. There are many examples of this, but one that sticks out to me is having your voice message translated to text (speech-to-text).

4. Conversational AI – a software agent can engage in a conversation with humans. You can think of Bots which are usually a first layer of contact when calling customer service.

5. Computer Vision – the computer has the ability to see and understand pictures and videos which it processes. Some abilities of computer vision are image classification and object detection. An example is uploading a picture of a dog and having it detect and name the type of object in the picture, which in this case is a dog.

What are AI Insights?

Since you now known the definition of AI and the five basic elements you may be wondering what is AI Insights, when did this feature become available, and what exactly do these AI Insights features in Power BI do?

The AI Insights group made its way into the general collection of features in June 2020. It can be defined as a collection of pre-trained AI functions which can be utilized to improve your data transformation process. There are 3 functions listed in the AI Insights groups: text analytics, vision, and Azure Machine Learning. Text Analytics and Vision functions are pre-built models in Cognitive Services. While Azure Machine Learning, allows data analysts to connect to data models in Azure.

Tip: You can hover over the top two AI Insights feature for a brief description of each element.

What is Cognitive Services?

Since you now know about AI Insights, we can now dive into Cognitive Services. Cognitive services are cloud-based services in Azure and provides developers with pre-built models so that they do not have to create their own. Creating your own models can be time consuming and requires a deep learning curve, and most developers and analysts are not data scientists, so having pre-built models saves a lot of time. Cognitive services have REST APIs and client library SDKs that help build AI into your apps. There are five pillars in cognitive services: Vision, Speech, Language, Decision, and Search. As you know by now, Power BI has two of the five. You can call these pre-built models by using an API. In Power BI, we call them through a function.

Who has access to AI Insights?

This is a very important part in the blog. I do not want anyone to try the AI Insights feature and become quickly disappointed with a message which reads ‘you need Power BI Premium.’ So let’s have a brief overview of which plans can access AI Insights capabilities.

You can access Cognitive services (Text Analytics and Vision) through the Power BI plans: Premium and Premium Per User (PPU). You can also use Cognitive services in Power Query with the Pro plan, but the method about doing this is different from Premium or PPU Plans. To use it with the Pro Plan, you will need to create a function that calls Cognitive services. In this function, you will need to provide a Cognitive Services key and endpoint from Azure. You will need to be familiar with M and JSON but if you are not, luckily there is a pre-defined template https://docs.microsoft.com/en-us/azure/cognitive-services/text-analytics/tutorials/tutorial-power-bi... .

The table below describes which plans include the AI feature capability.

Description of AI Functions in Power BI

With the Computer Vision Service, as of now there is only one function called tag images. According to Microsoft documentation, the tag images function returns tags on more than 2,000 recognizable objects. Some of these objects (not limited to) are humans, scenery, and objects.

Text Analytics helps us gain meaning from written text. Some functions are: Sentiment Analysis, Key Phrase Extraction, and Language Detection. A favorite among people seems to be sentiment analysis. I really enjoy Key Phrase Extraction.

Machine Learning, as mentioned above, uses algorithms to teach a computer model to make better decisions from data. To consume Machine Learning models, you or a data scientist will need to train and deploy a machine learning model. You will also need a subscription to Azure and be given read access to see the model in Power BI. You can create Machine Learning models in Azure by using: Code, Machine Learning Designer, or Automated ML.

Conclusion

We have reviewed the Artificial Intelligence Elements. Define the defined AI Insights, and what these Insights do in Power BI. We have looked at Cognitive Services and whom has access to it by reviewing the plan types in Power BI.

In my next blogs, I will cover in more detail the capabilities and features of the AI elements.

I hope this has helped increase your knowledge about AI Insights. Please feel free to leave me feedback! I would love to hear your thoughts and opinions. 

To find out more, please visit itdatadiva.com or you can follow me on Twitter @itdatadiva or on LinkedIn https://www.linkedin.com/in/itdatadiva/

Thank you for reading my blog,

Priscilla