Data & Analytics STL

Titans of Society: How Planetary Science Helped Sandeep Singh Deliver Data Science Value to Clients

Sandeep Singh

When Sandeep Singh was a child, he wanted to be an astronaut.

As he grew older, he learned it wasn’t so much that he loved “sailing into space” as it was physics, mathematics and the science of using data to draw insights. His passion was rooted in discovery – understanding the mechanics of how things worked.

And so he became a data scientist, working first with planetary constraints and then with societal.

In 2004, when the Cassini spacecraft arrived at Saturn and started collecting data, it became the hotspot for scientists to discover new theories. In 2011, Sandeep was fortunate enough to collaborate on a team that simulated the atmosphere of Titan, Saturn’s largest moon.

Cassini would send back hyperspectral data where sunlight is reflected and collected by VIMS (Visual Infrared Mapping Spectrometer) to Earth, providing insights that enabled the team to map the surface of Titan.

Imagine a 64×64-pixel picture (64 pixels on the x-axis, 64 on the y-) where each pixel represents a square kilometer of the surface. The spectral bands are recorded on the z-axis at a particular spectral resolution. Each pixel would have absorbed some light, so when these pixels are displayed on a color image, each is a different color depending on whether the absorption had been from water, methane or acetylene. This helps reduce the complexity of the z-axis, which shows the depth of light absorption, giving Sandeep reduced three-dimensional data to analyze.

Once Sandeep had the data formatted into columns and attributes, he could analyze it using traditional techniques, like building unmixed linear models. 

This is similar to what Sandeep does as a data scientist leader at Daugherty Business Solutions. Except, instead of extraterrestrial compounds, it’s human behavior and societal effects.

To make a prediction, Sandeep determines multiple dependent variables that are driven by a business problem. But he typically has hundreds to thousands of variables that might not correlate 1:1 with what he’s trying to predict. For example, whether a customer used a Visa or Discover card might not connect with why umbrella sales went up in the spring. To reduce the number of dimensions in that model, it makes sense to combine profiles into one (in the above case, a generic term named Credit Cards).

With planetary data, Sandeep worked with variability that is different from day-to-day data. The data took years to acquire, so the complexity of it (or noise) had to do with collecting it, as the distance between Earth and Titan is too far to gather enough quality data for a good signal-to-noise ratio (the amount of good data to the amount of noise). Although Sandeep had to do a lot of correction, the vision and scope remained the same throughout – to discover new compounds.

In the corporate world, the data is much more rapid compared to planetary data, so it has a high rate of turnover. It also produces an infinite number of economic variables and moving targets. The vision and scope are directly tied to the business problem, which tends to change, so the scope for discovery fluctuates along with the vision.

As a result, planetary data requires a traditional scientific bottom-up approach to the hypothesis (If Titan’s atmosphere is similar to Earth’s, it can give us insight into how life formed on Earth). In contrast, societal data requires top-down directives to set parameters (How do we increase sales by 3 percent?).

Ultimately, analyzing data would be useless if it didn’t serve a purpose. The Cassini spacecraft had been sent to Saturn and Titan specifically because Titan closely resembles prebiotic Earth. Sandeep had a roadmap to work with – what the Earth looks like today, what Titan looks like today, and a question to explore: “What would it take for Titan to get to where Earth is today?”

“We don’t know what started life on Earth – what the conditions were,” Sandeep said. “But Titan might give us clues about those conditions.”

Similarly, at Daugherty, clients ask questions such as: 

  • “Is the innovation we’re thinking about even possible?”
  • “How do we ensure the data science models created are integrated and leveraged across the whole organization?”
  • “How do we institutionalize innovation that delivers business value?”
  • “What are the best practices to prioritize and implement data science into an organization?”

Just as Titan might give us a clue to Earth’s prebiotic conditions, accurate data gives us insights into these questions and helps clients adapt new models for innovation.

This is similar to Daugherty’s model for excellence with clients. When we engage with a client, we focus on top-notch quality because we want a long-standing relationship – one that spans decades. As a result, we focus on where the client currently is and where they want to be – and then we roadmap the path to get there.

This is what drives Sandeep to become a leader in the data space at Daugherty. He wants to help clients transform their products and portfolios using data. Many people in the digital space need support understanding how the data-driven world works. Sandeep’s passion is to educate them and spark their curiosity.

Human nature, after all, is driven by curiosity. If we weren’t curious, we wouldn’t have explored space in the first place. Keeping that curiosity alive is how Sandeep thrives as he works within a complex world with complex problems.

Do you have any unique quirks you’d like to share with the enterprise? Email us at