Art Tennick’s Post

Plotly Studio - there are tables, and then there are Plotly Studio Tables natural language prompt only required to build what you see in the video prompt shown below I am not using the Plotly Studio Data Table visual that it always builds for you I wanted to create another version to experiment and compare, so I typed in a few words to generate the Python to generate my new table The source data contains only Internet Sales and Reseller Sales as measures/metrics Plotly Studio kindly created most of the other numeric columns for me I asked only for the I:R Ratio column to be added - without supplying any formula (no Python, no SQL, no M, no DAX, no Excel formula) not shown in the video is sorting on multiple columns - to do that, hold down the Shift key as you click column headers my prompt: 'Create a data table component using Dash Ag-Grid. sort Month in chronological order header background colour dark blue filter icon in white add dropdown to group by Country, Month, Product, Year - default to Month add a column to show ratio of Internet Sales to Reseller Sales and call it I:R Ratio light blue background and bold black font for Total Sales make all columns 160 pixels wide even-numbered rows in light green hover row colour orange cell selection box with broad black border make card 1230 pixels wide make grid 1200 pixels wide in Page Size dropdown use 20, 50, and 100 rows with a default of 20 make the grid 800 pixels high' PS and I have only covered a fraction of what is possible - I didn't include user input boxes, colour map conditional formatting, spark lines, graphics in cells, style sheets, drill-down, exporting from table, and a whole lot more - you can also use an HTML table rather than AG:Grid which gives access to a wide range of features here is the AG:Grid documentation - some of the very advanced features are Dash Enterprise only and may not work in Plotly Studio - and difficult to write prompts, so you end up hacking the Python and deploying to Dash Enterprise rather than to Plotly Cloud (I need to check and verify this later): https://lnkd.in/e2kHZ5MD Eszter Kovacs Brian Julius #plotlystudio #python

Plotly Studio - template for those who prefer the dark no coding, no formulas, no GUI point-and-click nothing required except pointing Plotly Studio to your data source it uses AI to generate prompts based on your data each prompt generated creates a written specification and writes quite a bit of Python - all powered by the in-built AI and the Python generates all your visuals - just sit and watch your dashboard appear not even time to grab a coffee as it takes about 2 or 3 minutes, and you end up with maybe 8 to 10 visuals arranged in a dashboard you can tweak the generated prompts - that's how I got dark mode (prompt below) and/or create your own prompts in natural language - each prompt you write results in a new visual I chose the Theme object and looked at its prompt - it was one line I added a second line to request dark mode everywhere Here is the finished prompt: 'Professional business analytics theme emphasizing data clarity and insights, clean modern typography, and subtle grid lines for precise data reading Use the plotly_dark template' the first line had been written by AI - I added the second line (not a lot of work!) there are other named templates apart from plotly_dark ("plotly", "plotly_white", "ggplot2", "seaborn", "simple_white", "none") plotly is the default, unless you name a different template or you can create your own templates PS I also added some words to the the AI-generated prompt for the Layout object, specifically to change the Hero section, which is the very top of the dashboard here is one line I added: 'Header Hero background with a gradient of black, blue, light blue, very light blue, light blue, blue, and black colors' Brian Julius Eszter Kovacs #plotlystudio #python

I Built GridScript: A New Way to Transform Excel Data with JavaScript and Python TL;DR GridScript.io lets you open Excel/CSV files, run JavaScript or Python scripts to transform the data, and export the result — all in your browser. No backend. No install. Instant feedback. Developers constantly bounce between spreadsheets and code when working with data — cleaning CSVs, merging reports, converting formats. I wanted a way to stay inside the browser and use the languages I already know (JavaScript and Python) to transform and visualize data. That idea became Gridscript.io. But it's also more than that, Gridscript offers access to Tensorflow.js and Scikt-learn allowing engineers to directly build AI/ML models. Next.js – fast frontend framework AG Grid – Excel-like grid for editing and filtering Monaco Editor – same editor used in VS Code Client-side architecture – no data leaves your browser Undo/Redo system – safe experimentation Create a New Pipeline Upload an Excel, JSON or CSV file Write a few lines of code in JS or Python Instantly see the grid update Export https://lnkd.in/gU-TAens

I Built GridScript: A New Way to Transform Excel Data with JavaScript and Python TL;DR GridScript.io lets you open Excel/CSV files, run JavaScript or Python scripts to transform the data, and export the result — all in your browser. No backend. No install. Instant feedback. Developers constantly bounce between spreadsheets and code when working with data — cleaning CSVs, merging reports, converting formats. I wanted a way to stay inside the browser and use the languages I already know (JavaScript and Python) to transform and visualize data. That idea became Gridscript.io. But it's also more than that, Gridscript offers access to Tensorflow.js and Scikt-learn allowing engineers to directly build AI/ML models. Next.js – fast frontend framework AG Grid – Excel-like grid for editing and filtering Monaco Editor – same editor used in VS Code Client-side architecture – no data leaves your browser Undo/Redo system – safe experimentation Create a New Pipeline Upload an Excel, JSON or CSV file Write a few lines of code in JS or Python Instantly see the grid update Export https://lnkd.in/gU-TAens

You Want to Be a Data Master? It’s about commanding data like it owes you answers. Python: The Power Player This is where data dreams are built. Why it matters: Python runs the show in analytics, automation, and modeling. Example: Predicting churn? Clean with pandas, model with scikit-learn, visualize with matplotlib. All in one language. SQL: The Language of Data You can’t master data if you can’t access it. Why it matters: Every warehouse speaks SQL. You don’t just query data—you interrogate it. Example: Grouping, filtering, or finding trends directly from millions of rows before the dashboard even loads. JavaScript: The Collector Data doesn’t just appear—it’s captured. Why it matters: JavaScript powers everything you track on the web. Example: One line in GTM tells your analytics tool that someone filled out a form, watched a video, or clicked “Buy.” No JS, no insights. Period. The Winning Combo: Python + SQL + JavaScript This trio covers everything: • JavaScript captures data • SQL structures it • Python turns it into meaning it’s storytelling through data. Learn one language, then learn how they talk to each other. Ready to master the stack that runs the modern data world? Drop your favorite coding language below and tell me what it’s done for your data game. #DataAnalytics #Python #SQL #JavaScript #RLanguage #DataEngineering #CareerGrowth #AnalyticsCommunity #GTM #GA4

🚀 Introducing My Python Data Analysis Software! I’m excited to share one of my latest projects — a Python-powered Data Analysis Software built with Tkinter, Pandas, and Matplotlib 🎯 This tool was designed to make data analysis easy, visual, and interactive, even for beginners. 💡 Key Features Include: ✅ Load and preview CSV or Excel files ✅ Display dataset summary (rows, columns, missing values, etc.) ✅ Generate descriptive statistics instantly ✅ View correlation matrix to detect relationships between variables ✅ Visualize any column (histograms, bar charts, etc.) ✅ Clean and modern UI/UX In the screenshot below, you can see how the app displays the distribution of age and generates automatic statistical summaries — all with just a few clicks! 📊 Building this project helped me strengthen my Python, GUI design, and data visualization skills. 💬 I’d love to hear your thoughts — what feature would you like me to add next? #Python #DataAnalysis #Tkinter #Matplotlib #Pandas #DataScience #Project #Programming #FatoluPeter #Portfolio

Got a great question during today's Discord Python + AI office hours! Q: How do you recommend extending the RAG repo (azure-search-openai-demo) with another tool, like to query a database? A: The repo currently uses only a single tool, searching an AI search index. I see two main approaches for adding new data sources or tools: 1️⃣ Router LLM pattern The main model routes between available tools — for example, calling search_ai_search or a new query_database function — and then passes results to an “answer” model. ✅ Predictable ✅ Easy to debug Drawbacks: ⚠️ Harder to make multiple tool calls for the same question ⚠️ No iteration or reflection 2️⃣ Agent pattern (looping calls) Instead of routing once, you let an agent loop — calling tools as needed until it’s done. ✅ Great for reflection and chaining multiple tools Drawbacks: ⚠️ Agent might call tools too often (or not at all) ⚠️ Adds latency, cost, and less control So if you just want one new tool call (like a SQL query), the router pattern is fine — just register your new tool and adjust the schema. If you intentionally want a multi-step reasoning workflow, an agent loop is the way to go, and there you likely want to use Microsoft agent-framework or another AI agent framework. Join us for OH every week in the Discord for more discussions like this! http://aka.ms/aipython/oh

Excel vs SQL vs Python One Task, Three Mindsets I built this quick visual guide to show something powerful: The same data task loading, filtering, or analyzing takes on a whole new identity depending on the tool you use. Here’s the story behind it 👇 🔹 Excel is where intuition lives — drag, drop, and visualize. It’s fast, familiar, and perfect for quick insights. 🔹 SQL is structure and control — clean queries, clear logic, and scalable data handling. 🔹 Python (Pandas) is freedom — automate, customize, and let your code tell a repeatable story. What’s fascinating is that the logic never changes, only the language does. Once you understand the thinking behind data not just the syntax you can move seamlessly from spreadsheets to scripts. This table isn’t just a comparison; it’s a reminder that true data fluency means being bilingual (or even trilingual) in how we work with information. Which one do you find yourself using the most lately Excel, SQL, or Python? #DataAnalytics #Excel #SQL #Python #Pandas #DataScience #AnalyticsTools #CareerGrowth #DataStorytelling

Most popular Python libraries for data visualization: Matplotlib – The most fundamental library for static charts. Best for basic visualizations like line, bar, and scatter plots. Highly customizable but requires more coding. Seaborn – Built on Matplotlib, it simplifies statistical data visualization with beautiful defaults. Ideal for correlation heatmaps, categorical plots, and distribution analysis. Plotly – Best for interactive visualizations with zooming, hovering, and real-time updates. Great for dashboards, web applications, and 3D plotting. Bokeh – Designed for interactive and web-based visualizations. Excellent for handling large datasets, streaming data, and integrating with Flask/Django. Altair – A declarative library that makes complex statistical plots easy with minimal code. Best for quick and clean data exploration. For static charts, start with Matplotlib or Seaborn. If you need interactivity, use Plotly or Bokeh. For quick EDA, Altair is a great choice. Hope it helps :) #python

Why I Never Begin with a Tool , I Always Begin with a Question I never launch Python, Power BI, or Excel before starting a new data project. I begin with the question, which is even more crucial. "What issue are we attempting to resolve?" "What conclusion will this data support?" "What information is truly necessary for the team to know?" Even the best tools cannot save the project if this step is skipped. Dashboards that are aesthetically pleasing but serve no purpose are the result. However, everything else falls into place once the question is clear: Cleaning gets simpler. Analysis narrows down. Meaningful insights Visuals become straightforward. Decision-making speeds up Clarity is produced by questions, not by tools. If your team is collecting data without clear questions behind it, this is the week to fix that before more time and energy gets wasted. Let’s talk.