The two row samples above (15 rows each, from `raw.AI_Usage_and_Impact_on_Students_and_Professionals`) show the dataset's overall shape and its quirks. I've left `Monthly_Income` out of my discussion because it is person-identifying. **What a typical row looks like** - Each row is one surveyed user (`User_ID` like U0207), either a Student or a Professional. - Rows span 21 countries, including the Philippines, Canada, the US, Pakistan, Germany, India and the UK. - Each row records one AI tool (ChatGPT, Claude, Copilot, Gemini, Perplexity, Meta AI) and one purpose (Coding, Studying, Writing, Customer Support, Content Creation, Research, Data Analysis). - Each row also has a free-text task description, usage hours, monthly AI cost, and three ratings: productivity (1.6–10.0, mean about 5.23), accuracy (mean about 3.43) and satisfaction (mean about 6.15). - It also has time saved per week, work or study hours per day, and a Yes/No/Maybe recommendation. **Data quality issues visible in the sample** - **Duplicates:** the 300 rows cover only 288 distinct `User_ID`s. U0207 appears in both samples, for example. - **Inconsistent gender labels:** there are 12 variants, including Male, male, MALE, M, Female, F, FEMALE and blank (10 rows). - **Inconsistent tool names:** "Chat GPT", "chatgpt" and "ChatGPT" all appear, as do "copilot" and "Microsoft Copilot". - **Inconsistent education labels:** there are 17 variants, for instance "undergrad" next to "Undergraduate". - **Country casing:** "france", "india" and "canada" appear in lowercase. - **Numbers stored as text with units:** values like "16 yrs", "1.6 hrs", "$20 " and "6.5 hrs" sit in columns that should be numeric. - **Missing values:** productivity is missing in 16 rows, accuracy in 15 and satisfaction in 7. Some rows also lack a tool, purpose, task text or profession. - **Mixed languages:** some professions and tasks are in German, French or Portuguese, such as "Kundenbetreuerin mit KI-Unterstützung". Analyses by gender, tool or education level should clean these labels first, or the groups will be split across spelling variants.