When it comes to big data and how companies use it in their internal operations, one phrase comes to mind: "vocational irony." While it may be obscure, its meaning is so common that it's become cliché in narrative fiction. The barefoot children of a cobbler. The broke accountant. The lovesick relationship expert. Or, more aptly, the technology company that provides cutting edge financial models for their customers while their internal machinery remains firmly rooted in the 20th century.
Earlier this year, global management consulting firm McKinsey & Company released a study on big data analytics that highlighted this point. The study featured insights into the rise of ecosystems and the coming wave of artificial intelligence. For me, one passage stood out:
"While investments in analytics are booming, many companies aren't seeing the ROI they expected. They struggle to move from employing analytics in a few successful use cases to scaling it across the enterprise, embedding it in organizational culture and everyday decision-making."
So how do we ensure big data and artificial intelligence improves our internal operational processes?
Big data is transforming nearly every single industry in the world. The housing finance industry is no different. Freddie Mac has been at the forefront of this transformation, introducing Loan Advisor Suite® in 2016, which put our risk assessment tools in our customers' hands to simplify the mortgage process and reduce costs. Loan Product Advisor®, the Suite's automated underwriting system, assesses hundreds of thousands of unique single-family loan files and appraisals monthly. And Loan Product Advisor continues to improve. We've recently added automated assessments of borrowers without credit scores, immediate collateral representations and warranty relief, and our automated collateral evaluation (ACE), which allows certain loans to be originated without an appraisal.
These are just some recent examples of how Freddie Mac is using data and advanced analytics in new ways to look at the mortgage experience differently, making smarter decisions faster through enhanced intelligence. We're helping create a better housing finance system. Our mission is to continue to do that – and do it better each time. But it is important that we make sure that we aren't walking around barefoot while we accomplish our mission.
There is an enormous opportunity to embed artificial intelligence (AI) and machine learning into a company's internal operational processes. Previously, companies had to make tradeoffs with their time, money, workforce and – importantly – their computing power. And inevitably, improving the customer’s experience was at the top of the list.
But with the availability of cheaper, scalable cloud computing, we can extend big data and AI in-house. It just makes sense to apply this cutting-edge approach to internal operations, so we can get more done, cut costs, and improve the speed of service. Not only can the cobbler's children have shoes, they can have the best shoes in town.
Since the early days, we've used expert rule-based AI systems for our customers, and we're moving into the next frontier. More specifically, that involves artificial intelligence for artificial intelligence.
Freddie Mac recently partnered with machine-learning startup DataRobot to help our data scientists find fast and more efficient methods for modeling our historical data. DataRobot's platform allows users to build and deploy highly accurate machine learning models in a fraction of the time it takes using traditional data science methods.
For Freddie Mac, this partnership is already allowing us to explore new machine learning algorithms in fast and powerful ways. Consider the following.
In the past, if we were trying to improve a predictive model, we would:
This cycle typically took many, many months.
Over the last decade, thousands of new algorithms have been created and new data sets have been introduced. In the old way of doing things, you would have a human use their best judgement to pick the correct data set and the correct algorithm.
What DataRobot provides is the ability to rapidly analyze and experiment with any structured data and then select the most likely algorithms that will give you what you're trying to predict from it. It doesn't even need to know anything about the data. It scans and profiles it using unsupervised machine learning, then selects dozens of algorithms to run on the data. In true Darwinian fashion, it creates a leaderboard that identifies the fittest algorithms to use. It can even evaluate combinations of algorithms to help our data scientists get the best answer.
That's artificial intelligence (algorithm) for artificial intelligence (algorithm selection).
From a customer perspective, it will help us improve our models faster and allow us to experiment more with non-traditional data that we believe will help improve the mortgage origination process. That's the cobbler. But what about the children?
At a time when we're all drowning in email and data, how do we filter out the noise, decide what's important and focus on it? Think about all the data-centric activities where a person must apply judgment and decide to: approve an invoice, flag a suspicious transaction or identify suspicious activity on your internal network.
AI and machine-learning can help our employees make faster, better decisions on what might be an outlier, what we need to pay attention to, and what might be our next best step. Now is the time to do it. We're at a time in history when it just makes sense. And it's part of our transformation into a better company, building a better housing finance system and reimagining the mortgage experience.
Forget about shoes, we've graduated to hover boots.