Machine Learning Overview
What is Machine Learning?
Machine Learning (ML) is a specific form of artificial intelligence that adapts and improves through experience and exposure to new information. Rather than programming software to take an action based on a stimulus and then manually adapting that program for each new stimuli, machine learning models adjust course upon encountering new information, based on data encountered in the past.
One everyday example: Streaming service recommendations. Most streaming services recommend content based on users’ interests. As you consume more and more content over time on one of these platforms, the recommendations shift.
Machine learning is a huge part of involve.ai’s magic, and we use a combination of the below best-in-class models and techniques to deliver predictions, insights, and recommended actions with over 90% accuracy.
involve.ai’s Machine Learning models
involve.ai’s 7- and 9-Factor models are linear regressors that take specific types of quantitative data that involve.ai has found to have the most impact on health (for example, product usage and interaction frequency), identify targets for each of those inputs (for example, “Two meetings per week is the ideal for this customer”), and synthesize it all in order to make predictions about customers’ likelihood to renew.
Named Entity Recognition (NER)
NER processes unstructured text (such as emails, meeting transcripts, or written comments) by recognizing and attributing meaning to specific phrases and words. By doing so, involve.ai’s models can bring attention to the most relevant components of customers’ emails or phone calls and suggest appropriate actions to take in response.
OpenAI’s Pre-Trained Transformer (GPT), another form of natural language processing, generates text with a warm, human-like tone. involve.ai uses this technology in the Smart Summary feature that summarizes interactions for end users, and we are working on expanding its use in other exciting ways - stay tuned!
AI can be used to analyze written or spoken text and to determine whether the person who wrote or spoke the content was expressing positive, negative, or neutral feelings. Involve.ai uses two types of sentiment analysis: TextBlob and NER + Polarity. These forms of AI provide you with, respectively, insight into the sentiment of a call or email exchange and the sentiment of a support ticket.
Clustering models sort your data into cohorts, telling you in very specific ways whether certain attributes of customers make them likelier to behave in similar ways. For example, through the use of two different clustering models, involve.ai can help you understand whether your customers’ tenure affects the impact of frequent in-person meetings, or whether a particular industry has a higher need for onboarding support.