Michael Salmon

I’d love to open a candid dialogue with fellow nonprofit Analysts and Data Managers about data collection, inspired by an insightful post this morning by Andy McMahon. Andy pinpoints a key issue: The modeling features we can create from nonprofit data assets often lack the predictive power needed to generate reliable, useful outcomes. Why not? 🤔 For starters, philanthropic motivations are complex and highly individualized. There's a broad range of influences on a person’s decision to give. Modeling all of them for your entire constituent base is impractical. Realistically, the resources needed to create such comprehensive data assets are beyond most of our means. However, we should continue to ask donors about their motivations, documenting them in contact reports for future reference. While these specifics may not enhance our ability to model motivation across the whole database, they're invaluable for fundraisers preparing for meetings. Gathering personal motivation is typically a slow, one-to-one process of asking for it, so what if we wanted to model who would be most willing to share it? This leads me to a somewhat debatable point: Nonprofits could reasonably access more behavioral data to support fundraising. Example: How might you identify the top 25 acquisition prospects who are most likely to take a discovery meeting? Non-Donors likely have sparse records in your CRM—no donations, minimal personal info. Maybe a wealth rating? What are predictors of someone's willingness to meet? How about their reactions to previous mass communications. It's likely that your organization has been reaching out to most of these prospects via phone, email, texts, and videos for years. How did they interact with that content? Where is that data? All too often, content engagement data remains in third-party tools, only some of which is integrated into our CRMs when deemed critical. Yet there's a wealth of information in living those tools about how constituents interacted with those touches. Video watch rates, email shares, call-to-action clicks, phonathon call lengths...etc. The potential insights from engagement with mass communications are extensive. But....where is that data? 🤷‍♂️

12

26 Comments

"Data is a precious thing and will last longer than the systems themselves." — Tim Berners-Lee In the data and analytics space, we often focus on the latest tools and technologies. However, the true value lies in the data itself and the insights we can extract from it. As professionals in this field, we must bridge the gap between technical innovation and business impact. Let’s continue to push the boundaries of what’s possible, but never lose sight of the business problems we aim to solve. Our work should always translate into actionable insights and tangible results. To business leaders and other professional across organisations, embrace the power of data-driven decision-making. Invest in understanding the potential of analytics and data science to transform your strategies and operations. Collaboration between technical and business teams is crucial for harnessing the full potential of our data assets. Together, just like my team at Turner & Townsend and I, let’s build a future where data not only unlocks some progress but creates lasting value. How are you leveraging data to make a difference today? #DataScience #Analytics #Leadership #BusinessIntelligence #DataDriven

32

1 Comment

As a data professional, I am always looking for ways to read the story of data or to tell the story of data in the most effective way. Like everything, in order to tell an effective story in data visualization we also need to have a strategy in place. It is not just about fancy charts or filling up a dashboard page. Here is something I want to share with those who are just beginning this journey of data visualization. There are 5 important elements of visualization. I am attaching a link that I found. it explains each of the below elements really well. #data #visualization, Data Visualization Society hierarchy. Location Size Gradient Color Shape

21

3 Comments

Saw this article making the rounds and it did not disappoint. It is absolute gold. I'd encourage anyone interested in statistics, research, or data informed decision making in general to read the whole thing. And then read it again. Here's a taste of the premise, but it goes into much more detail of what you can do instead: • Don’t base your conclusions solely on whether an association or effect was found to be “statistically significant” (i.e., the p-value passed some arbitrary threshold such as p<0.05). • Don’t believe that an association or effect exists just because it was statistically significant. • Don’t believe that an association or effect is absent just because it was not statistically significant. • Don’t believe that your p-value gives the probability that chance alone produced the observed association or effect or the probability that your test hypothesis is true. • Don’t conclude anything about scientific or practical importance based on statistical significance (or lack thereof) "In sum, 'statistically significant'—don’t say it and don’t use it." #statistics #statisticalsignificance https://lnkd.in/g9G2huyD

2

I've been experimenting with RStudio's Copilot integration, and it's scary from two perspectives. - It's scarily good at integrating with both R script files and RMarkdown/Quarto. I like it. - It has terrifying and terrible implementation from a data security standpoint. There isn't a built in way to say "only enable this in a new project if I've manually chosen to do so". Let me walk you through the problem: You're working on "Personal Project 42" and have Copilot enabled. You receive an email from a client asking them to take a look at something, so you open "Client 343" project in a new RStudio window and open one of the scripts. Boom. That whole script file has gone through GitHub's Copilot. This would be less of an issue if RStudio has personas, but it doesn't. It feels like if I really want to be secure with GenAI I need to use completely different user accounts on my mac between personal and consultancy work.

18

5 Comments

I have just submitted my reproduction of Cara Thompson's work, #BarBeRight, to the plotnine contest. Here is the link to the tutorial: https://lnkd.in/gBtktGsR Credit goes to Cara Thompson for her insightful tutorials and superb work on Data Visualization. #plotnine #datascience #rstats #python

25

2 Comments

🚀 Exploring USA State Minimum Wage Trends Over Time 📊 I've been programming in R for a long time. I am really enjoying going into my old code and asking Github Copilot (https://lnkd.in/edsfWJ3J) to optimize the code. Here is a refreshed script I wrote a couple of years ago which highlights the changes in the minimum wage across different states in the USA from 2000 to the present. This project utilizes data from the Federal Reserve Economic Data (FRED) and the U.S. Department of Labor, and leverages R libraries such as fredr, tidyverse, geofacet, ggrepel, and ggtext. Key Insights: - Variability Across States: The minimum wage trends show significant variability across states, with some states experiencing more frequent and higher increases compared to others. - Federal Minimum Wage as a Baseline: The dashed line represents the federal minimum wage, which has remained at $7.25 since 2009, serving as a baseline for many states. - Geographical Trends: States on the West Coast, such as California and Washington, have seen substantial increases, reflecting policy decisions aimed at improving living standards. Feel free to check out the code on my GitHub Gist: https://lnkd.in/dxTKgiZB #DataScience #DataVisualization #Economics #MinimumWage #RStats #FRED #LaborEconomics #PolicyAnalysis

19

4 Comments

Check out Orr Group's recent #AI training, which covers everything you need to know to craft effective prompts and elicit quality responses from AI large language models like ChatGPT and Google Bard. Watch the recording to enhance your skills in leveraging these powerful tools effectively. #AItraining

2

A great review of mine and Adnan Masood, PhD. book ‘Responsible AI in the Enterprise’. This is all becoming increasingly relevant as AI regulation becomes clearer globally and Enterprises are increasingly shaping up how they risk assess and govern AI. ‘Responsible AI in the Enterprise by Adnan Masood, PhD. and Heather Dawe was a great treatment of all the things that can go wrong in AI / ML projects, important principles in the design and operation of these systems, and ways of detecting and mitigating fairness issues in AI Systems.’ Thank you Matt Eland! https://lnkd.in/eWefyccD #responsibleAI #AIEthics #AI

14

🎇🎇🎇🎇🎇🎇🎇 Season 10 finale of the PolicyViz Podcast with Nancy Organ and her new book, Data Visualization for People of All Ages. We talk about a range of issues relating to data visualization education and literacy. ➡️Importance of Data Visualization. Nancy underscores the necessity of data visualization skills in the modern world and how these skills can be nurtured from a young age. Nancy’s book aims to demystify complex data concepts, starting with basic data units and building up to more sophisticated visual representations. ➡️Educational Approach. Her book includes self-assessment tools and classroom exercises to facilitate learning with an emphasis on making the content relatable and straightforward for a broad audience. ➡️Ethics in Data Visualization. There is a strong focus in Nancy’s book and in our conversation on maintaining data integrity and making ethical choices in visual storytelling. ➡️Techniques and Tools. We also discuss various data visualization formats such as timelines and flowcharts and how to understand different data encodings to enhance perceptibility and engagement. ➡️Incorporating Visualization into Education. Finally, Nancy and I talk about integrating data visualization and data science into the K-12 curriculum and how to promote visual thinking across multiple disciplines and learning environments. Thanks for tuning in over this season! I hope you learned a lot about creating and communicating your data and analysis. Be on the lookout for great new episodes coming this fall. In the meantime, enjoy your summer!!!! 🎙Listen: https://lnkd.in/edYPM8kF 📺Watch: https://lnkd.in/ezvb65mZ

11

Just had an interesting call with a colleague that used to query voter file data back when it was stored on a big round tape. It made me wonder: Has constituent data become so numerous and complex that accessing and querying it has become too inaccessible for those who need it most? (I think so.) How can we overcome this and give the power back to people to query their data without the need of data professionals? Can we leverage an AI semantic layer to "google" our own data? Can we really use more tech to solve tech problems? If I work myself out of a job, is it a job well-done?

5

1 Comment

The vastness of #bigdata and #realworlddata can be daunting. Designing studies to minimize bias, either when estimating causal effects (e.g. https://lnkd.in/g4XzMSTB) or correlations, are often discussed and, if designed properly, will result in outcomes that could benefit patients. However, the key to unlocking the true potential of #bigdata always starts with: ▶ Are you asking the right question? ▶ Has your question been answered already? ▶ Have you assured good provenance for your data source(s)? Get in touch with our team at 📣 DiscoverNOW@imperialcollegehealthpartners.com, whether you are looking to be in a better position at the start of these questions, or if you are in need of study design support or #realworlddata.

3

LangSmith Evaluations: RAG Evaluation (Answer Correctness) Evaluations can accelerate LLM app development, but it can be challenging to get started. We've kicked off a new video series focused on evaluations in LangSmith. https://lnkd.in/gw9b7nNU https://lnkd.in/gTyvUtcz

4

It’s #DataViz #SuperBowl season. If you see me posting images of Rick Klein or Steve Kornecki- know I’m not pushing electoral discussion- but I’m #nerding out on mainstream #news data visuals on dashboards. This is a side of me I have to surpress on social media because it gets political REAL FAST in the comments- but I just want to talk about the COMMENTARY and Live stats PRESENTATION y’all! And the maps!!! #datacommunicator True story, when drafting and pitching my #datacommunicator role profile / job description research- I had to make an extremely advanced version as well as a very basic draft of the position before our internal review folks understood what I was talking about. The extremely advanced version was for public speakers on the news like Steve Kornecki, who has gone on to do live data viz presentations at other events like the Kentucky Derby. It’s why I remain patient to see what will come from #VR and #AR layering, as promise has been shown to get to a future Hans Rosling would “squee!” about as #tech and #datascience evolves. #nerdalert Anyone else this data viz nerdy?

5

Really like this comprehensive and straight forward guide to set up dashboards and present data for internal business use.

10

As a Data Scientist, I know that telling a technology story is never easy. We often focus too much on the technology itself and overlook the culture and business it’s intertwined with. I watched *SIGHT* during this long weekend break. I highly recommend it to anyone interested in technology, innovation, and storytelling. The film offers a thought-provoking perspective and is surprisingly captivating and entertaining throughout. I could sense the passion, intelligence, diligence, and sense of humor in the film. The philosophical depth brought by director and screenwriter Andrew Hyatt in telling Dr. Ming Wang's story is truly remarkable! Spanning history, culture, and political systems, the story is both touching and inspiring. It's a must-watch for professionals looking to gain insights and broaden their understanding of these important themes.

38

❓Is anyone aware of a good public dataset with came_from/went_to columns in a many-to-many relationship? Preferably dozens of each, with the same came_from also being possible went_tos. If there is a category grouper associated with them, that would be even better. I want to create a public version of a solution I came with at work that combines the from/to in a single list relative to any given selected went_to so you get a list of all its upstream sources and all its downstream destinations. The work example is post acute care, which you could enter from any of various facilities from acute to post-acute to home and depart while going to any of various facilities from acute to post-acute to home - so any given facility has a set of upstream sources and downstream destinations. As an example, Superstore would not be a good for this (and it is not a good fit for Sankey despite most Sankeys on Tableau Public using Superstore as a data source). Please put links to any suggested datasets in the comments. 🙏 Thanks in advance!

7

11 Comments

I contend that Polars is the future of data analysis. As I continue to explore this wonderful package, I am amazed and inspired by its alignment with the dplyr package. I delved into data filtering and selection in Part 2 of Learning Polars the Right Way. This article highlighted various ways of filtering data. Please share my work with your network and leave a comment on how to improve future articles or topics you want me to write about. #pydata #datascience #python #deeplearning #rstats

10

This is going to be a fun session! Talking about some of the new features of Power BI, so awesome visualizations, and my favorite dataset gymnastics. Come learn the benefits of turning your data on its head (literally)! Join us by registering for the monthly lunch & learn here: https://hubs.la/Q02vZcxX0

4