Mining useful information from a database can be difficult, especially when your database is large and structured in a complex fashion. Yet information is like lifeblood for businesses, and those who can extract it can use it to gain market share and dominate competitors. I talked with Adam Carrigan, an entrepreneur and former management consultant at Deloitte who is also the co-founder and COO of MindsDB. Adam graciously allowed me to mine some information on how to use artificial intelligence (AI) to extract information from a database. and I’m sharing my conversation for readers whose businesses can benefit from Adam’s insights and extensive experience.
MITCH: Why do so many companies have trouble extracting useful information from their databases?
ADAM: While most companies have large amounts of data, either the data is not structured in a way that makes it useful or they do not have the tools needed to extract information efficiently. Machine learning and predictive analytics are a great solution to these problems. Unfortunately, these tools generally required specialized data science resources and knowledge. Typically, these high-demand data professionals work for large enterprises with the budget to successfully adopt the latest in technology.
Even for large enterprises, data science projects can be complicated and time-consuming. Since the actual data users need to go through data scientists to extract information, these deployment pipelines can become significantly backed up.
MITCH: Explain the concept of predictive analytics as it applies to machine learning.
ADAM: Predictive analytics existed well before machine learning. It simply involves analyzing historical data to detect any useful patterns to help generate future predictions and forecasts. On the other hand, machine learning is a subset of artificial intelligence, which creates algorithms that learn from a series of data inputs and outputs to be able to perform classifications, predictions, and more when given the input. I wrote a concise article on the difference between AI, machine learning, and deep learning.
Ultimately you can think of machine learning as one way to perform predictive analytics, and in many cases, it is the most accurate. But it is not the only way. For example, if you used Excel and its simple forecast formula to predict a trend, this is predictive analytics, albeit probably not a very useful or accurate one.
MITCH: How does adding AI functionality to existing databases bring companies benefits?
ADAM: Embedding AI capabilities into an existing database make the power of machine learning accessible to anyone able to run an SQL query. As noted earlier, larger enterprises were the only organizations with the technical resources and expert personnel to implement machine learning projects successfully; this is now no longer the case.
Simply adding AI to an existing database opens up the data’s full potential by bringing machine learning directly to the source. Organizations can now glean actionable predictive insights from their data with a simple database query because predictions now look, feel, and act like regular tables in a database, queryable just like any other. This concept is called AI-Tables. The ultimate benefit of using AI-Tables is that machine learning models become less complicated and costly and can be developed and scaled quickly.
MITCH: How does one go about adding an AI to a database?
ADAM: Adding AI to an existing database is a relatively simple process for those databases that allow external tables; this feature goes by different names depending on the database. However, databases that currently support external tables and thus also support AI-Tables include MariaDB, MySQL, PostgreSQL, and ClickHouse. However, expect more databases to provide this functionality as the technology grows in popularity.
The user installs the open-source software onto a cloud, on-prem or virtual server. Note the importance of ensuring the server’s GPU supports any specialized instruction sets required for successful operation. Additionally, the server instance needs to provide enough RAM to run efficiently; this will vary depending on your data’s size and complexity.
Next, just provide the database credentials, and you are ready to start building and using machine learning models.
MITCH: What is the MindsDB approach to making machine learning accessible?
ADAM: As noted earlier, building a machine learning model is complex and requires specialized resources. It’s an expensive and time-consuming process. Even with an accurate model, moving it into production is a tedious process.
However, MindsDB now provides the revolutionary AI-Tables for most databases. The MindsDB solution allows users to easily create, train, and deploy ML models with only basic SQL skills. MindsDB also reduces the complexity of model training by leveraging an innovative Explainable AI feature at the source of the data.
Ultimately, models are developed quickly and deployed at speed and scale with AI-Tables. This approach enables businesses of any size to promptly derive actionable predictive insights from their data. The power of machine learning is no longer the exclusive domain of data scientists and senior software engineers.
The benefits to businesses are numerous: predicting churn, detecting fraud, and optimizing a customer’s lifetime value; all by using basic SQL queries. Just connect it to your database, run a query, train the model, and get forecasts as tables in your existing databases. Notably, MindsDB customers have saved up to 60 percent in costs by using AI-Tables.
MITCH: What do you think the future holds for predictive analytics and machine learning? How will this affect commerce and society, as well as the impact on the consumer and end-user?
ADAM: In the end, simplifying machine learning and bringing it to the database allows data users to reap more benefits of predictive analytics. Giving machine learning tools to the end-user will reveal exciting new uses for the technology and new ways to leverage the insights gained. At that point, expect the expanded use of machine learning to grow at an even higher rate. As machine learning matures as a technology, companies will enjoy an improved ability to predict customer behavior. This enables them to design products and promotions to truly provide what consumers want.
MITCH: Thank you, Adam, for sharing these insights with our TechGenix readers.
ADAM: You’re welcome.
Featured image: Shutterstock