Power BI – Simplifying the complex world of artificial intelligence

The What

Customer satisfaction and the customer experience are crucial aspects of every business, not just for the business themselves, in understanding how to improve, but also for influencing the customer in which product to buy.  It is therefore crucial for you to listen to and understand your customer feedback.  This feedback will help you understand trends in your business and help you gain insights into what your customers are saying about your industry and your competitors.

Think of an e-commence website as an example.  You will often see a rating allocated to a product based on individual reviews by the customer. How often have these ratings influenced your decision-making process and determined whether you buy the product or not.  You may wonder how these scores are determined, if they have any validity and if it is something that can be replicated.

Manually collecting and analysing data can be extremely time consuming and inaccurate due to subjectivity. Instead, using built-in features of Power BI like AI Insight, and Sentiment Analysis, you can now automate the management, and analysis of customer reviews.  Power BI text analytics helps streamline the process through the automation of data collection and analysis. You can obtain this data in a matter of seconds, allowing you to gain insights into your customer satisfaction in real time.

The How

The Text Analytics service provides natural language processing, which is the action of harnessing computer processing power to read and analyse large amounts of language.  Given raw unstructured text, it can extract the most important phrases, analyse sentiment, and identify well-known entities.  The tool enables you to quickly see what your customers are talking about and how they feel about it.

• Score sentiment

The Text Analytics associated with the AI Insights feature of Power BI allows its users to build a sentiment analysis model. It utilizes Azure Cognitive Services to obtain the sentiment score.  Sentiment Analysis considers a text input, such as a customer review, and runs a machine learning algorithm which assigns a score ranging from 0 to 1. A score of 0 indicates negative sentiment and 1 indicates positive sentiment. The model in cognitive services is trained with a dictionary of texts that are mapped to its corresponding sentiments.

Using whatever processes necessary to obtain your customer reviews, you simply need to collate them into a data table.  Once you have this source table available, the next step is to load this into Power BI and apply sentiment analysis over the raw text contents.

Note: To use these AI features you will need a Premium Capacity on your workspace

All you need to do is link it up to the text field within your data set that you wish to analyse, then hit ‘ok’.

Once processed, Power BI provides the results of the sentiment analysis as an additional column alongside the original data.  By building a calculated measure on this sentiment analysis score you could then classify the results into the 5-star ratings that is common throughout the retail industry. Very happy (>=0.8), happy (>=0.6), average (>=0.4), unhappy (>=0.2), very unhappy (<0.2).  Go one step further by adding geographical details about your customers and you are well on the way to creating insight that will amaze.

• Key Phrases

After you have calculated a rating from how your customers feel about your product, it is important to understand why they feel the way they do. Power BI has thought of this too.  Using the Extract Key Phrases feature allows you to obtain more detail about customer responses by extracting commonly used words. The Key Phrase Extraction function evaluates unstructured text, and for each text field, returns a list of key phrases. The function requires a text field as input. This process allows you to analyse the main topics provided in customer reviews and is particularly useful in today’s customer-centric world because it allows you to read between the lines of your customer reviews and to gain better understanding.

Presenting the key phrases is made simple too with Power BI’s word cloud visual.  While traditional charts and visualization techniques are great at making sense of structured data, they generally fall short when it comes to textual data. You often have to look for creative new ways for presenting text data. Word cloud is a visual representation of word frequency and value. Use it to get instant insight into the most important terms in your data.

Note: Key phrase extraction works best when you give it bigger chunks of text to work on. This is opposite from sentiment analysis, which performs better on smaller blocks of text. To get the best results from both operations, consider restructuring the inputs accordingly.

• Detect Language

A third option within the Text Analytics feature is around language.  If you sell and ship goods all over the world, then it’s imperative that you understand all your customers and can interpret their feedback.  The language detection function evaluates text input, and for each field, returns the language name and ISO identifier. This function is useful for data columns that collect arbitrary text, where language is unknown. The function expects data in text format as input.  Text Analytics recognizes up to 120 languages.

Conclusion

Artificial Intelligence has been making its presence felt in various domains. The value derived from it has been making a positively disruptive impact in many industries. AI is now slowly getting embedded into self-service BI tools. With no exception to this trend, Power BI is one of the early adopters for AI and Machine Learning Capabilities. The need for this capability and insight is only going to increase as our shopping habits migrate to be more and more online.  Customer feedback and reviews are and will become more and more important as the demand for sales satisfaction continues to climb.  With the ease in which these text analytics services can be utilised it is highly recommend that you get onboard and start acting on their outputs now.

Author

---

Richard Storey, Senior Solution Architect

Richard is a Senior Solution Architect in Capgemini, with a specialism in architecting best practice solutions and delivering enterprise reporting solution.  He has extensive knowledge of visualisation tools, primarily Power BI, but also including the likes of Tableau, Qlik Sense, Salesforce Einstein Analytics and Microstrategy. ​​