Technology Tales

Top YouTube Channels for Data Science

KDnuggets compiled a ranking of the top 15 YouTube channels covering data science content, determined by searching the platform using the term "data science", scraping the top 100 channel results, removing those without publicly available subscriber counts and re-sorting by subscriber numbers. The ranking also incorporates total view counts and views per subscriber to give a fuller picture of each channel's reach and engagement.

The list is led by Edureka!, which boasts over 2.4 million subscribers and more than 197 million views, followed by Joma Tech and Simplilearn. StatQuest with Josh Starmer focuses on breaking down statistical and machine learning concepts into digestible steps, whilst Ken Jee combines data science with sports analytics.

Channels such as Data School and 365 Data Science cater to those seeking structured learning paths, and Andreas Kretz focuses specifically on data engineering. The list is capped at 15 entries on the basis that channels beyond that threshold offered diminishing relevance to the data science field.

Apache Arrow is a columnar memory format designed for efficient data interchange and in-memory analytics, enabling fast access to structured and nested data across modern computing hardware. It supports zero-copy reads to eliminate serialisation overhead and is implemented through libraries available in multiple programming languages, facilitating high-performance analytics and integration with various tools. The project is community-driven, emphasising open collaboration and consensus-based decision-making, with contributions from diverse organisations and individuals.

Conda is an open-source tool for package, dependency and environment management that supports multiple programming languages. It can be installed via two main distributions: Miniconda, an Anaconda-preconfigured installer, and Miniforge, which is maintained by the Conda-forge community and preconfigured for the Conda-forge channel, with both available through Homebrew. Documentation covers the essentials for new users, including environment creation and management, alongside a full command reference, configuration guidance, cheat sheets and a glossary. Those wishing to contribute to the project can find guides covering project governance, contribution processes and development environment setup.

rOpenSci Packages: Development, Maintenance, and Peer Review

The rOpenSci Dev Guide provides comprehensive guidance for individuals involved in the development, maintenance and peer review of R packages within the rOpenSci ecosystem. It outlines best practices for creating and testing packages, and explains the peer review process including the roles and responsibilities of authors, reviewers and editors.

The guide also offers strategies for sustaining packages after onboarding, covering collaboration, documentation, promotion and the use of GitHub as a development platform. Templates and resources are provided to support various stages of package management and contribution, with an emphasis throughout on structured workflows and community engagement.

rOpenSci promotes open and reproducible research by developing shared software tools and fostering collaboration among researchers and engineers through the R programming language. It employs a peer review process to validate and improve scientific software, supporting a community of developers and maintainers who contribute to packages across disciplines such as statistics, data visualisation and geospatial analysis.

Initiatives such as the Champions Program encourage leadership in open science, whilst projects such as R-Universe provide platforms for discovering and publishing R tools. Efforts to expand accessibility include translating documentation into multiple languages and creating resources for open-source contributions. The organisation collaborates with institutions and supports workflows that enhance data analysis, security and reproducibility in scientific research, reflecting a broader commitment to making rigorous, transparent research practices more widely achievable.

Open-Source Portal for Clinical Study Evaluations

This open-source portal has been developed to serve as a centralised collection of links, programmes and scripts related to clinical study evaluations, addressing the difficulty researchers face when trying to identify what open-source solutions exist in this space. The portal includes metadata storage and user-friendly search and navigation tools, with content gathered manually and supplemented through header analysis and available metadata. It has grown over time to include tools such as DaVinci, carver, oak and teal, as well as other R packages, conference videos, podcasts and an education section. Plans include enabling user ratings and edits, expanding available content, and developing a process for managing and downloading open-source SAS macros in a manner comparable to how R packages are installed.

Visual Define-XML Editor

The VDE Dataset Viewer is a multiplatform tool, available on Windows, Linux and macOS, built to help users visualise and explore datasets in several formats, including Dataset-JSON v1.1, NDJSON, compressed Dataset-JSON, XPORT v5 and SAS7BDAT. It supports features such as reading large datasets, filtering with value autocomplete, sorting, row and column navigation, cell selection, metadata information and automatic updates. Users can customise their viewing experience through settings covering numeric date formats, number rounding, dynamic cell height, automatic width estimation and encoding control. The tool also offers API access based on the DataExchange-DatasetJson-API specification and is released under the MIT Licence.

Business Science University provides online education focused on applying data science and machine learning in business contexts, delivered through virtual workshops that guide learners through problem-solving processes from data analysis to deploying interactive applications. The programme targets data analysts, consultants and students, offering courses that cover real-world applications in areas such as customer analytics, financial modelling and text processing, with an emphasis on systematic project workflows and advanced techniques such as automated machine learning.

Learners gain practical skills through end-to-end projects, develop web-based tools for organisational use and build portfolios on GitHub, supported by a lifetime access model and certification upon course completion. The curriculum is structured into tracks that progressively build expertise, though the university clarifies that it does not offer accredited degrees.

Smart Submission Dataset Viewer

The Smart Submission Dataset Viewer is an open-source application designed to inspect and analyse CDISC SDTM, SEND and ADaM submission files formatted in the modern CDISC Dataset-JSON 1.1 standard. It leverages contemporary technologies such as RESTful web services and CDISC CORE for validation, offering advanced capabilities beyond traditional tools used by regulatory authorities, which often rely on outdated formats like XPT. The tool is particularly beneficial for regulatory reviewers, pharmaceutical sponsors, contract research organisations and technology providers involved in CDISC electronic submissions, and it is among the first open-source implementations of the CDISC Library API.

What is eCOA, and How Does it Improve Clinical Trial Data Quality?

Electronic Clinical Outcome Assessments (eCOA) have emerged as a significant development in clinical trials, using electronic devices such as smartphones and tablets to collect patient-reported outcomes and other critical data more efficiently and accurately than traditional paper-based methods. By enabling real-time data monitoring, reducing errors and improving patient compliance, eCOA enhances the reliability of trial results and supports timely interventions.

The technology is particularly valuable in scenarios requiring precise data collection, such as monitoring suicidal ideation, managing complex assessments or conducting large-scale studies, and offers additional benefits including cost savings and seamless integration with other medical technologies. It encompasses various forms, including patient-reported (ePRO), clinician-reported (eClinRO), observer-reported (eObsRO) and performance-based (ePerfO) outcomes, each tailored to specific trial needs.

Selecting an eCOA solution involves evaluating a provider's scientific expertise, operational experience and technological capabilities to ensure alignment with the trial's objectives and regulatory requirements. As clinical research continues to evolve, eCOA plays an increasingly important role in advancing data quality and streamlining processes across the pharmaceutical industry.