Harish Chatla’s Post

THIS PASSES CODE REVIEW. STILL WRONG. Look at the query in the image. It tries to calculate daily active users by picking the latest event per user per day. At first glance, everything looks correct: • Window function • Proper partitioning • Clean aggregation But there’s a subtle issue. In real systems, events don’t always arrive in order. Late-arriving data can completely change what “latest” means. The query runs successfully. The numbers look reasonable. But the result can be wrong. This is exactly how incorrect metrics end up in dashboards without anyone noticing. Your challenge: What is wrong with this query? How would you fix it? Drop your answer in the comments. Follow Harish Chatla for real-world SQL challenges. Repost if this could help someone preparing for interviews or debugging production queries. Subscribe on YouTube for full breakdowns. #SQL #DataEngineering #AnalyticsEngineering #BigData #DataAnalytics #LearnSQL #DataQuality #ETL #DataEngineeringLife #DataRejected

Day 60 of #100DaysOfSQLSolved 💻 Solved “Weather Observation Station 10” on HackerRank ✅ 🔍 Problem: Retrieve unique city names that do NOT start with vowels (a, e, i, o, u). 🧠 What I practiced: Using DISTINCT to eliminate duplicates Applying string filtering with LEFT() / SUBSTR() Writing efficient WHERE conditions Strengthening logical filtering in SQL 💡 Key Insight: Excluding specific patterns is just as important as selecting them—this improves real-world data filtering skills. 📈 Consistently improving my SQL skills step by step towards becoming a Data Analyst. #SQL #DataAnalytics #HackerRank #100DaysOfCode #LearningJourney #FutureDataAnalyst 🚀

🗓️ SQL Challenge Day #32: Consecutive Numbers 🔹 Find numbers appearing 3+ times in a row! 🔢 🔹 Problem:   Identify values with ≥3 consecutive occurrences:   ✅ Consecutive = sequential `id` values   ✅ Return distinct numbers only  🔹 Solution (Window Function Approach):  SELECT distinct num AS ConsecutiveNums FROM (     SELECT         num,         id,         LEAD(id) OVER (PARTITION BY num ORDER BY id) AS l1,         LEAD(id, 2) OVER (PARTITION BY num ORDER BY id) AS l2     FROM Logs ) t WHERE l1 - id = 1   AND l2 - l1 = 1; ✅ Result: Accepted  💡 Key Takeaway:   **LEAD() checks sequence gaps!** By comparing current `id` with next two `id`s:   - `l1 - id = 1` → next record is immediate successor   - `l2 - l1 = 1` → third record is also consecutive   ⚠️ Why partition by `num`? Ensures we only compare same numbers!  👇 Your turn:   Have you used LAG/LEAD for detecting sequences in logs or time-series data? What patterns did you find?  #SQL #LeetCode #DataEngineering #ProblemSolving #Coding #LearningInPublic #Database #DataAnalytics

There's a SQL function that fits in one line. No subquery. No CASE WHEN. No workaround. Most analysts have never typed it. It's called NULLIF(). Here are the problems it solves: 1) Your query crashes with a division-by-zero error. The usual fix? A clunky CASE WHEN. But using NULLIF() - one clean line. 2) Converting 'N/A' to NULLS if 'N/A' is stored as a string in the raw data. 3) Treating empty strings as actual NULLs. Why does nobody use it? Because CASE WHEN works. And when something works, we stop looking. But NULLIF() is cleaner, shorter, and easier to read at a glance - especially inside larger queries where CASE WHEN adds 4 extra lines of noise. Every time you write CASE WHEN, there's a NULLIF() waiting to replace it. Had you heard of NULLIF() before this? 👇 #SQL #DataAnalytics #DataEngineering #SQLTips #BusinessIntelligence #DataScience

don't wrt query 🚀 Day 34 – Top 50 SQL Questions Today’s problem: Product Price at a Given Date 🔍 Objective: Find the price of each product on a specific date. 💡 Approach: Used a correlated subquery to get the latest price before the given date Handled missing data using default values Focused on retrieving time-based records efficiently 📈 Key Insight: Selecting the most recent valid record is a key pattern in SQL interview questions. #SQL #DataAnalytics #LeetCode #ProblemSolving #Day34

Most SQL problems are not about syntax; they’re about recognizing patterns. For example, detecting consecutive user activity, or streaks, isn’t a “SQL function” problem — it’s a pattern recognition problem. Here’s a simple approach to tackle it: - Assign sequence using ROW_NUMBER - Normalize gaps using date – sequence - Group to identify streaks Advanced SQL is about transforming data into shapes where the answer becomes obvious. Once I understood this, problems like consecutive logins, sessionization, and retention cohorts started to feel much easier. What SQL pattern took you the longest to understand? #SQL #DataEngineering #Analytics #ReLearningInPublic

This SQL query runs perfectly. No errors. Clean logic. Looks correct. But the result is completely wrong. Look at the query in the image. It tries to calculate total revenue per customer. And at first glance, everything seems fine. But there’s a subtle mistake in the aggregation logic that completely distorts the output. This is exactly how production dashboards get incorrect numbers — not because queries fail, but because they silently return misleading results. Your challenge: What is wrong with this query? Write the correct SQL in the comments. Follow Data Rejected for real-world SQL challenges. Repost if this could help someone preparing for interviews or debugging production queries. Subscribe on YouTube for full SQL breakdowns. #SQL #DataEngineering #AnalyticsEngineering #DataAnalytics #Database #LearnSQL #SQLTips #TechCareers #QueryOptimization #BusinessIntelligence #DataRejected

🚀 I spent hours debugging my SQL… turns out the query wasn’t the problem. The logic was. Everything ran fine. No errors. Clean output. But the result was still wrong. (took me longer than I’d like to admit to realize it) That’s when it hit me: 👉 Writing correct SQL ≠ producing correct metrics Most mistakes don’t break queries — they quietly distort the outcome. Since then, I’ve started questioning the logic more than the code. Curious — what’s a mistake that changed how you approach data? #DataAnalytics #SQL #DataEngineering #AnalyticsEngineering #DataQuality #BusinessIntelligence #LearningInPublic

🚀 Day 43/100 – LeetCode SQL Challenge Today’s problem: “Customer Who Visited but Did Not Make Any Transactions” 🔍 What I learned: How to use LEFT JOIN to keep all records from one table Understanding how NULL values help identify missing data Using WHERE t.visit_id IS NULL to find visits with no transactions Applying GROUP BY + COUNT() to calculate results per customer 💡 Key Insight: When the question says “no data” / “missing” / “without” → think 👉 LEFT JOIN + IS NULL 🧠 Final SQL Approach: SELECT v.customer_id, COUNT(*) AS count_no_trans FROM Visits v LEFT JOIN Transactions t ON v.visit_id = t.visit_id WHERE t.visit_id IS NULL GROUP BY v.customer_id; 📈 Takeaway: This problem strengthened my understanding of joins and filtering logic—very useful for real-world data analysis and interviews. #Day43 #100DaysOfCode #SQL #LeetCode #LearningJourney #DataAnalytics

🧹 Day 14: Clean Data is Happy Data! Today’s SQL session was all about String Manipulation. In the real world, text data is rarely perfect. It has extra spaces, inconsistent casing, or needs to be combined for better reporting. My toolkit for today: UPPER/LOWER: Standardizing categories for consistent filtering. TRIM: Removing those pesky leading and trailing spaces that break queries. CONCAT: Merging columns (like Product + Category) into readable strings. REPLACE: Bulk-updating text patterns (e.g., changing "phone" to "device"). SUBSTRING/LEFT/RIGHT: Extracting specific parts of a string for shorter identifiers. Formatting might seem small, but it’s the difference between a messy database and a professional report! 📊 #SQL #DataCleaning #DataAnalytics #100DaysOfCode #DatabaseManagement #TechSkills