Articles
Notes on data without theory for its own sake
I write about situations where a weak definition, a poor model or bad data quality changes the conclusion. The terminology matters less than the decision it affects.
Short weekly notes on data in the Telegram channel (in Ukrainian): @belyakovconsult
· 4 min readWhy last month's report changedLate records and status changes quietly rewrite past reports. How to agree a cut-off date and show revisions honestly.
· 6 min readWhich BI platform to choose: a ten-criteria comparison mapPower BI, Tableau, Qlik, Looker and SAP Analytics Cloud in one interactive map: from the main strength to what most often makes the work harder.
· 4 min readDocument type can matter more than the number itselfWhy a document number alone is not enough for reliable identification in data exchange, and what to add so a matching number does not become an error.
· 4 min readOne person, three records: register duplicatesWhy one person shows up in a register several times, what it does to reports and payments, and where to fix duplicates.
· 4 min readBI should not be the place where source systems are repairedWhy cleaning and stitching data inside Power BI makes reports fragile, and which part of that work belongs in the data model.
· 2 min read98% complete does not mean 98% qualityWhy a high completeness rate says little about whether registry data is usable, and which checks reveal its real quality.
· 4 min readData quality dashboard: don't hide the problem in an averageFour indicators on the first screen, a matrix instead of one number, a target set by the data owner and a PDF report. Plus a demo on synthetic data.
· 5 min readThree levels of data quality in registersWhat ISO 8000-8 says about syntactic, semantic and pragmatic quality, and why registers need all three.
· 4 min readYou can see data governance when numbers conflictTwo teams bring different figures, and both are right. What that moment reveals about data governance and the four questions to start with.
