Bas Dohmen (How to Power BI)

A Power BI common request from business users is to do weekly calculations. This can become tricky, especially if there are anomalies in the cadence of reporting dates for the data. In this example for the 30-year mortgage rate data from FRED, they typically published the data on Friday in the early 1980s (currently they do it on Thursday), but that wasn't always the case. During this illustrated time period in the fall of 1981, they occasionally reported the results on other days, like a Wednesday or a Thursday. We can't use the DATEADD function here as a filter to move the time intelligence ruler back on a seven day interval to determine the previous week's rate because it's not always exactly seven days earlier. One solution for this problem is to tap into the new Enhanced DAX Time Intelligence options that came out in the September Power BI updates last week. It allows us to do week-over-week and week-to-date running totals using built-in DAX functions. This feature is currently in preview, however, and we also need to configure the calendar options in a specific way for it to work. It's one approach that can save us a lot of potential headaches, even if it takes more work to set up. If you work with modeling and analysis on a weekly basis, I definitely recommend checking out these new enhanced DAX time intelligence functions when you get a chance. #PowerBI #TimeIntelligence #AlwaysBeLearning

213

5 Kommentare

Enterprise-Ready Data Architecture: 18 Proven Levers for Data Quality Transformation   1.    Embed Data Governance with Clear Ownership Assign domain-specific data stewards (e.g., Finance, Sales) accountable for the quality of their datasets. Use lightweight governance (e.g., quarterly reviews) to avoid bureaucracy while enforcing accountability. Case Insight: Capital One embedded data stewardship into business lines for performance ownership. 2.    Automate Validation at Ingestion Points Block low-quality data with real-time validation (format checks, null thresholds, deduplication). Example: Reject CRM entries with missing customer IDs at source to protect downstream analytics. 3.    Implement Data Lineage Tracking Use tools (e.g., Collibra, DataHub) to visualize data flow from source to consumption. Rapid root cause analysis prevents recurring issues. 4.    Proactive Monitoring with Threshold Alerts Monitor key indicators (e.g., data freshness, record volume drop) with real-time alerts. Route anomalies to engineering within defined response windows (e.g., <1 hour). Example: BNY Mellon flags sharp data volume anomalies in trade data feeds within 15 minutes. 5.    Standardize Critical Metrics Enterprise-Wide Define and enforce one source of truth for core metrics (e.g., Revenue, CAC, MAU). Insight: Avoid cross-department confusion by having unified metric logic definitions. 6.    Build Dedicated Data Cleansing Pipelines Run automated cleanup tasks (e.g., deduplication, address standardization) for high-priority domains. Prioritize high-value data (e.g., customer, financials) over operational logs. 7.    Version Control for Datasets & Pipelines Use tools (e.g., Delta Lake, DVC) to snapshot datasets and pipelines. Enable rollback when errors are detected in transformations or model training. 8.    Enforce Cross-Functional SLAs Create SLAs for pipeline reliability and quality checks (e.g., fix within 4 hours, validate within 2). Practice: Penalize repeated offenders to ensure accountability. 9.    Incremental Quality Testing Test quality at each stage (ingestion → transform → consumption) rather than only end-to-end. Example: Validate sensor data formats before entering transformation layer. 10. Root-Cause Analysis with Blameless Post-Mortems Post-incident reviews should focus on fixing process gaps, not assigning blame. Action: Document and share learnings in a central wiki to prevent recurrence. 11. Invest in Observability Tooling Use modern tools (e.g., Monte Carlo, Anomalo) to track data drift, schema changes, freshness anomalies. Use Case: GE uses data observability to maintain integrity across distributed manufacturing systems.   Continue in 1st and 2nd comments.   Transform Partner – Your Strategic Champion for Digital Transformation   Image Source: ABN AMRO

113

11 Kommentare

🚨 Did you know most Power BI dashboards underperform not because of the data but because of poor design? At AlphaBOLD, we have seen how cluttered visuals, misaligned KPIs, and confusing layouts slow down decision-making instead of driving it. That is why we created our blog: Top Power BI Dashboard Design Tips for Clear and Impactful Visuals. Inside, we share: ✅ Common dashboard design pitfalls that frustrate executives ✅ 10 best practices for building dashboards that deliver insights quickly ✅ Why clear design is not cosmetic but business critical For leaders, the takeaway is simple: a well-designed dashboard is more than a reporting tool. It is a decision-making accelerator. 📊 Read the full article here: https://lnkd.in/dk2cHGFu

105

1 Kommentar

🔎 DAX Solutions for Outlier Detection in Power BI 💡 Learn how to spot and handle outliers in Power BI with practical DAX techniques, step-by-step guidance, and clear visualizations to keep your dashboards accurate and trustworthy. 🔗 https://lnkd.in/eSWb_Zd8 👉 For more: https://lnkd.in/e5WKaY2q

40

“We use SAC now — Power BI didn’t work.” ... 😏 Sure. And next year, we’ll use something else — because that one didn’t work either. 😉 The truth is: tools don’t fail. Strategies do. A new platform won’t fix missing processes, undefined data ownership, or lack of clarity in use cases. ➡️ Technology amplifies what’s already there — the good and the messy. Choose your tool wisely — but build your foundation first. #PowerBI #Fabric #Strategy

67

1 Kommentar

File format vs table format: what’s the difference? Our new blog breaks down how Iceberg and Parquet complement each other and how Qlik makes building an Iceberg-based lakehouse on Amazon S3 effortless. Read more: https://bit.ly/424tsdZ

55

1 Kommentar

New Blog Post! Continuing with #DAXUDF, I know all the rage is with model-independent functions in daxlib.org (which is great), but what about model-dependant functions? Like calculation groups, even when they have dependencies on the model, they are essentially the same function so it's a pitty we need to copy and paste around and fix manually those dependencies. Like I did several times for calc groups, I show how to *industrialize* functions that have model dependencies with a Tabular Editor #CSharp Script , and do so in a way that can then be used by the previous macro that generates measures based on the function adding all the necessary properties (name, expression, formatstring, displayfolder and table). Also I know the new Calendars are the new hot topic, but for now I decided to have 4 classic Standard Time Intel calculations from Daxpatterns.com into functions Anyway, here's the article with links to the scripts! I hope you like it https://lnkd.in/dGiWRsM8 As a teaser, just look at this beautiful video of what can you do:

74

11 Kommentare

Guess how many calculations you need to show WTD, MTD, QTD, or YTD... In today’s Intermediate Next-Level Tableau class, we built a calendar to verify the calcs. We created one flexible calculation: ✔ Week-to-date, month-to-date, quarter-to-date, year-to-date Then we created two calcs: ✔ Prior periods (like prior month-to-date) ✔ Same period in the prior year All controlled by a single parameter. Then we used map layers to design the calendar heatmap. All in one sheet! Here’s how it works 👇 What’s your favorite way to compare time periods in Tableau?

254

10 Kommentare

📌 Power BI Workflow – From Raw Data to Insights 🔗 Explore more data resources → https://lnkd.in/dR96YGaA Power BI is one of the most powerful tools for turning raw data into meaningful insights. Here’s the step-by-step workflow every analyst should know: 1. Get the data → Import from Excel, SQL Server, SharePoint, Web, APIs 2. Clean the data → Remove blanks, rename columns, set correct data types 3. Shape & transform → Merge queries, pivot/unpivot, group rows 4. Load the data → Apply changes into Power BI Desktop 5. Create relationships → Use keys, star schema, adjust directions 6. Build measures (DAX) → Totals, averages, KPIs, business logic 7. Design the report → Charts, slicers, cards, maps 8. Add interactivity → Filters, drill-through, slicers 9. Publish & share → Publish to Power BI Service, schedule refresh, share dashboards 🎓 Recommended courses to level up your data skills: IBM Data Science → https://lnkd.in/dCZvDFwF SQL Basics for Data Science → https://lnkd.in/dtZFSq2f Meta Data Analyst Professional Certificate → https://lnkd.in/dtcBsxQm Google IT Automation with Python → https://lnkd.in/ddvJ4y3d Generative AI for Data Scientists → https://lnkd.in/dramyTj4 #PowerBI #DataAnalysis #BusinessIntelligence #DataScience #ProgrammingValley

61

🚀 Confused about Power BI Storage Modes? Here’s a quick visual breakdown of the 4 core modes: → Import → DirectQuery → Dual → Direct Lake (exclusive to Microsoft Fabric) 🎯 Why does it matter? As a Data Analyst, your choice impacts: 📊 Report performance ⏱️ Data refresh strategies 🌐 Gateway dependencies ⚡ End-user experience Every mode determines how your data flows, updates, and performs in real-time. 💡 Pro Tip: Just because Direct Lake offers near real-time access, doesn't mean it's the right fit for every scenario. Choosing the right storage mode for your specific use case is critical — each one has unique advantages and limitations. 📌 Save this post for your next Power BI model decision! #PowerBI #DataAnalytics #MicrosoftFabric #BusinessIntelligence #StorageModes #DataModeling #CodeBasics #DataStrategy #PowerBITips #BItools #DataEngineer

30

2 Kommentare

Introducing AI and Agentic Development for Power BI: an article that explores concepts and scenarios for integrating AI and AI agents into BI development, with specific examples for Power BI. And if that is not enough, 80+ pages of a new whitepaper and a new video course are available on SQLBI+.

63

1 Kommentar

Context Transition in Power BI  Three similar-looking formulas. Three completely different mechanisms. Let me break it down 🛠️ 📅 Example 1: Best Selling Day We have an existing Calendar table in our data model. Since it's structured at the day level, MAXX iterates through each day, calculates [Total Sales] for that specific day, and returns the maximum. The granularity of the Calendar table determines that we get the best-selling day. 📦 Example 2: Best Selling Product Same logic, different table. Our Products table exists at the SKU level (product code level). MAXX iterates through each product, evaluates [Total Sales] for that product, and returns the winner. The table's granularity gives us the best-selling product. 🗺️ Example 3: Best Selling Region Here's where it gets interesting. We don't have a dedicated dimension table for regions in our data model. So we create a virtual table using VALUES(Sales[Region]). This extracts all unique regions from the Sales table, and MAXX can then iterate through them to find the best-selling region. 🔑 The key insight: Context transition happens when MAXX iterates and converts each row into a filter context for the measure to evaluate. But the granularity of what you're iterating over, whether it's a physical dimension table or a virtual table, determines what "best" actually means. Physical tables = leverage your data model structure  Virtual tables = create analysis paths that don't exist in your model #PowerBI #DAX

303

8 Kommentare