by Jelani Harper 25 November 2019
Rapid advancements in predictive and prescriptive analytics have seemingly surpassed the overall utility of descriptive analytics. But as we strive to determine what will happen, and to prepare accordingly using technologies like machine learning, it is easy to forget the main value proposition of descriptive analytics which, although less celebrated, continues to endure.
Descriptive analytics doesn’t reveal what might happen, what should happen, or what your plan of action should be. Instead, it illustrates something much more concrete—what actually did happen and, with the proper analysis, what to do to get the most advantageous outcome out of a situation.
Sentiment analysis is perhaps one of the most pervasive use cases for descriptive analytics today. Analyzing customer or even employee sentiment on platforms either directly involving, or designed to mimic, popular social media channels is crucial for understanding how best to engage people.
By leveraging various forms of text analytics, natural language technologies, and even machine learning, organizations can master sentiment analysis with descriptive analytics, demonstrating “what the most and least active areas of the business are, so that you can then start targeting those areas that are maybe the least active, and that need an extra push around engagement rather than just doing a mass engagement,” Ciara O’Keefe, VP of customer experience at StaffConnect, told AI Business. “You can go after the ones that aren’t engaging just yet.”
Sentiment analysis engagement
The distributed nature of the contemporary data landscape, which is increasingly reliant on cloud and mobile technologies, makes engagement a key priority for most organizations. Although engaging customers is a best practice for ensuring patronage, many organizations with diverse locations, national, or even international operations, are tasked with engaging their own employees to maximize productivity. Sentiment analysis can measure engagement in both cases and suggest what actions are needed to optimize engagement.
A common means of fostering engagement is to issue surveys, which rely on descriptive analytics to “get a full report that can be exported to show results in bar or pie chart form, depending on the type of question,” O’Keefe explained. Although such quantitative measures work well for closed-ended questions, variations of text analytics are required for more granular insight. One such text analytics method for open-ended responses involves “a word list that shows the most frequently mentioned terms in bigger letters, making it easier to see trends,” O’Keefe added.
When attempting to automate understanding of what was actually said about key terms, concepts, and entities extracted from text, natural language technologies are extremely useful. According to Lyndsee Manna, SVP at Arria NLG, Natural Language Processing is defined as “text to data,” while with Natural Language Understanding, organizations are able to “understand what was said.” Both of these technologies are applicable to text as well as speech, making them important for sentiment analysis. Sophisticated analytics mechanisms—for sentiment analysis, BI, and other internal enterprise uses—are able to rely on Natural Language Generation and what Manna termed Natural Language Queries. Whereas the latter allows users to seek answers from data in natural language, the former is critical for issuing natural language responses that “automate the knowledge process of a human, when they see these things in the data – this is what I do, what I say, how I describe it,” Manna said.
Text analytics variations
Text analytics is often bolstered by natural language capabilities and enhanced by machine learning approaches like deep neural networks, allowing users to see “what are the common themes that are coming out the comments,” O’Keefe said. Even without natural language technologies, machine learning, and formal, dedicated text analytics solutions that involve much more preparation time than those relying on cognitive computing, organizations can still perform sentiment analysis on social media-like engagement platforms with the measures O’Keefe referenced. Moreover, they can rely on manual methods to do the same work, by “downloading the CSV of the data with the comments,” O’Keefe added.
Regardless of which approach is utilized, sentiment analysis and different forms of text analytics ensure descriptive analytics remain highly relevant in a world in which predictive and prescriptive analytics are lauded. In many ways, descriptive analytics is more beneficial than its younger siblings, because it’s perfectly transparent, doesn’t involve bias, and is rooted in solid fact as opposed to prediction. Moreover, descriptive analytics enables organizations to consistently derive value from sentiment analysis via various forms of text analytics—some of which may very well utilize the predictive capacity of cognitive technologies like machine learning and natural language processing. This combination is critical for understanding how both consumers and employees are reacting to relevant products, services, or business policies. You could say this insight is required for improving them and optimizing productivity.
Jelani Harper is an editorial consultant servicing the information technology market, specializing in data-driven applications focused on semantic technologies, data governance and analytics.