The 15-row sample above mixes students and professionals across several countries, tools and purposes. It also shows that the raw data is messy. **What the sample covers** - **Users:** 8 students and 7 professionals, aged from 16 to 63. - **Countries:** the Philippines, Canada, the United States, Pakistan, Germany, Australia, India and the United Kingdom. - **Tools:** ChatGPT, Claude, Microsoft Copilot, Perplexity AI, Google Gemini and Meta AI. - **Purposes:** Content Creation, Coding, Studying, Research, Writing and Customer Support. - **Scores and costs:** Productivity_Score runs roughly 4 to 7, and Accuracy_Rating and Satisfaction_Score are small integers. Monthly_AI_Cost is mostly 0, 10, 20 or 30. - **Recommendation:** Would_Recommend is Yes, No or Maybe. **Data-quality issues visible in the sample** - **Inconsistent formats in text-typed numeric columns:** - Age appears as `16 yrs` (U0138). - Monthly_AI_Cost appears as `$20 ` (U0140). - Time_Saved_Hours_Per_Week appears as `6.5 hrs` (U0140). - AI_Usage_Hours_Per_Day appears as `1.6 hrs` (U0207). - These need cleaning before any averaging. - **Inconsistent category labels:** - Gender uses `F` as well as `Female` (U0098). - Education_Level uses `undergrad` as well as `Undergraduate` (U0087). - AI_Tool uses `copilot` as well as `Microsoft Copilot` (U0140). - **Missing values:** these are scattered across columns: - Profession, Tasks_Performed and Productivity_Score for U0138. - AI_Tool for U0096 and U0085. - Country for U0087. - Monthly_AI_Cost for U0082. - **Implausible combinations:** - Some ages sit oddly with their education level, such as a 16-year-old with a Master's (U0138) and a 17-year-old with a PhD (U0085). - A 20 for Work_or_Study_Hours_Per_Day (U0085) is also suspect. - **Mixed languages:** the Profession and Tasks_Performed text is occasionally in German (U0230). I left out Monthly_Income because it is person-identifying; it can be summarised in aggregate. I can also profile null rates or label variants across the full 300 rows, or break the data down by tool, purpose or user type.