Demystifying Data Science | Mr.Asitang Mishra | TEDxOakLawn

Summary

Asitang Mishra, a data scientist at NASA’s Jet Propulsion Laboratory (JPL), describes data science as a broad, evolving field focused less on a particular job title or tool and more on solving human problems with data and computers. Data scientists often combine computing, mathematics, and knowledge of a subject area. They may come from disciplines such as physics, engineering, economics, or psychology.

Key Ideas and Examples

Anomaly Detection at JPL

Spacecraft instruments generate large volumes of data. Mishra’s team built algorithms to monitor data patterns and flag possible anomalies for human operators, reducing the need for people to watch the data continuously.

Algorithms Are Only Part of the Work

Algorithms process data to produce predictions or results. Data scientists may select and tune existing algorithms rather than write them from scratch. Mishra estimates that this accounts for only about 20% of the work; defining the problem, preparing the data, and understanding it take much of the rest.

He also mentions efforts such as DARPA’s D3M project, which aims to automate algorithm selection and tuning.

Translating Human Problems into Computer Problems

A major skill is reframing a problem so it can be addressed computationally. As an example, Mishra describes NASA’s effort to fit a large space telescope into a smaller rocket by applying principles of origami.

Caring for Data

The work includes finding or combining suitable data, cleaning it, and understanding what its fields and columns represent. Poor-quality data can lead to poor predictions.

Communication and Storytelling

Data scientists need to explain their work in language people can understand and show how a tool supports people’s decisions. They should not assume that every system needs to fully automate or replace human judgment.

Mishra illustrates storytelling with NASA’s SMAP satellite, which measures soil moisture and can help with flood and drought prediction, as well as assessments of crop productivity.

Open Source and Collaboration

Mishra describes sharing code so others can use, modify, and build on it. He mentions JPL’s support for open-sourcing projects, a build-it-yourself rover modeled on a Mars rover, and collaboration with the Apache open-source community.

Career Outlook

Mishra argues that data science does not require one specific degree. Relevant problem-solving and computing skills can come from many fields. The talk cites reported growth in data-science roles and projected job openings, while emphasizing the availability of online learning resources.

Reviews, Guides, or Tutorials

None are presented. This is an explanatory talk rather than a product review or step-by-step tutorial.

Main Speaker and Source

Asitang Mishra, data scientist at NASA’s Jet Propulsion Laboratory, speaking at TEDxOakLawn.