This year marked Kiosk’s third year participating in Pydata events and the second year of the all-online, global format. As in years past, leading developers, data scientists, and business leaders showcased the cutting edge of data discovery, visualization and machine learning tools while maintaining a steady focus on the ethical implications of tools and processes.
There was a particular emphasis on the explainability of models. We can no longer accept the “black box” approach where we build models, trust their results and make decisions based on their output. Decision makers in business, government, the scientific community and the public at large demand to know, and rightfully so, how the algorithms and models are arriving at their decisions and determining their “fairness”. As artificial intelligence and machine learning are used to make exceedingly more impactful decisions and affect our lives in a growing variety of ways, the moral and analytical choices being made need to be transparent.
As with every year, we were excited to bring back several ideas and a couple of products that have already joined our product ecosystem.
Data Profiling
We were excited to join a presentation given by Jeremy Goodsitt and Austin Walters from Capital One covering the setup and use of their Data Profiler suite of tools. The tools can be used to scan files and database tables for a variety of sensitive information including email addresses, phone numbers, bank, credit card and social security numbers and much more.
We were able to get the product up and running during the talk and have already added it to our already robust suite of scanning and compliance tools. This gives us an enhanced ability to keep track of any sensitive data wherever it lives in our ecosystem and, just as importantly, prevent it from entering our systems unexpectedly in the first place.
Python Dashboarding Shootout and Showdown, and Why Interactive Data Visualization Matters for Data Science in Python
We also got a whirlwind tour of the rapidly evolving world of data visualization in Python with the Python Dashboarding Shootout and Showdown, as well as Nicolas Kruchten’s talk on Why Interactive Data Visualization Matters for Data Science in Python.
Robust and interactive data visualization and dashboarding platforms used to be out of reach of most entities outside of large corporations, universities or research institutions. Last week, developers and analysts were treated to a display of four different free and open source data visualizations platforms, each with their own particular niche. Tools that would have required weeks of consultation, install and setup and a significant financial outlay now spring to life in minutes without having to spend a dime.
At Kiosk, we employ an array of different dashboarding, reporting and other visualizations platforms and tools to share insights gleaned from our data with our customers and partners to help guide their understanding and decision making processes. We are excited to see the rapid pace of development in the last several years and look forward to incorporating some of the new tools by Plotly to enhance our reporting and analytical offerings and make our data more readily available and self-service.
There’s truly never been a more exciting time to be developing in the data science sector, and particular in Python. The quality and availability of tools has never been better and things are constantly evolving, and swiftly at that. We look forward to continuing our engagement with Numfocus, Pydata and the community of business partners and developers to continue to enhance the quality of our data and analytical offerings. By leveraging new learning in smart ways we are able to drive faster, more informed and better optimized decision making and ensure our clients get the most out of every dollar they invest in their partnership with Kiosk.