Nick Tarazona, MD’s Post

if you are using a language model to create a quantitative output, consider this instead: 1. classical analysis 2. asking the language model to make a likert-type output instead 3. a hybrid approach of 1 and 2: you can run classical analytics on the LLM output, giving you access to many of the quant measures you want, but altering the interpretation slightly - you need to mention your LLM's training set influence

Thematic analysis is a method for analyzing qualitative data that involves reading through a set of data and looking for patterns in the meaning of the data to find themes. It is an active process of reflexivity in which the researcher's subjective experience is at the center of making sense of the data. The screen shot of mutiple respondent groups analysed against primary and secondary themes in the software helps to narrow down, support 'general to specificity' in inductive research approach. https://lnkd.in/g-6D_yDj #research #PhD #masters #thematicanalysis #qualitativeresearch #EU #UK

Interesting paper by Emmanouil Tranos et al., proposing a novel methodology to identify economic clusters over time using archive data from the JISC UK Web Domain Dataset (subset of the Internet Archive). For validation, they looked at Shoreditch in East London, and found some interesting details! 📘 Read the full paper here: https://lnkd.in/etPH5qCw #Research #EconomicClusters #Tech #DataScience

I simply need to accept that the vast majority of what is published in Statistics and Data Science is not mathematically rigorous. While the symbol pushing is syntactically correct, the claims are frequently provably wrong. One of the thorns in my side is the shear number of authors that claim their algorithm du jour generates Lebesgue measures, translation invariant, in the strong topology of function spaces. This is a provably false claim becuase the unit ball is not compact in function spaces. If we were to constrain ourselves to only analyses that were rigorously defensible we would be far more parsimonious in the algorithms we use and the conclusions we reach.

I hope that my contribution to the literature with the study ‘USE OF ARTIFICIAL INTELLIGENCE AND AUDIT ANALYTICS IN INTERNAL AUDIT PROCESSES IN THE PUBLIC SECTOR’ will be a guide for researchers working on the subject.

Understanding Subarray vs. Subsequence in Data Structures Today, let's unravel the distinction between subarrays and subsequences in the realm of Data Structures and Algorithms (DSA)! These concepts play crucial roles in problem-solving and algorithmic optimization, yet they possess distinct characteristics. 🔍 Subarray: A subarray is a contiguous portion of an array, consisting of elements that appear consecutively in the original array. It maintains the relative order of elements and cannot skip any elements. 🔍 Subsequence: A subsequence, on the other hand, is a sequence of elements that may not necessarily be contiguous in the original array. It retains the relative order of elements but can skip elements in between. ⚙️ Key Differences: - Subarrays are contiguous, while subsequences can be non-contiguous. - Subarrays preserve the order of elements, while subsequences maintain the relative order but may skip elements. Understanding the nuances between subarrays and subsequences is pivotal in algorithm design and problem-solving strategies. Whether optimizing for efficiency or exploring creative solutions, clarity on these concepts is indispensable. Excited to delve deeper into these distinctions and apply them in tackling real-world problems in DSA! Join me on this journey of exploration and mastery. #DSA #Subarray #Subsequence #Algorithms #LearningJourney #gauravsah

More Is Less: https://lnkd.in/d9qwU4Yc This -underrated- paper proves (mathematically) that huge datasets contains arbitrary correlations, indepedently of the nature of the data. Results are based on: - Poincare recurrence: famous to provide regularities in large subsets, e.g Szemerédi's theorem - Ramsey Theory: famous to provide lower bound for a fixed "correlation" so that it holds in any set of that size or bigger.

RECODING QUANTITATIVE VARIABLES INTO QUALITATIVE ONES – TECHNIQUES AND THEIR PRACTICAL APPLICATIONS https://lnkd.in/d2JMnjEA

✨ Excited to share that our paper "Challenging Fairness: A Comprehensive Exploration of Bias in LLM-Based Recommendations" has been accepted to IEEE BigData 2024! 🎉 In this paper, we explore the fairness in large language model-based recommendation systems. We studied different interesting research questions and addressed them in terms of different fairness metrics, highlighting the urgent need for more equitable solutions in LLM-based recommendations. A version of the paper can be found here- https://lnkd.in/geUWCyBQ Thanks to my co-author Shahnewaz Karim Sakib. We look forward to discussing our findings and engaging with peers in the community to further explore these vital issues. See you at IEEE BigData! #IEEEBigData #FairnessInAI #MachineLearning #ArtificialIntelligence

Starting another course at Stanford University on statistics to gain comprehensive knowledge about data collection, organization, interpretation, analysis, and representation.