# Data Analysis > Source-anchored research for Data Analysis. Run status: COMPLETED. 4 facets · 11 facts · page: https://llms-explorer.com/tree/data-analysis/ ## Structure and components - Visualize data before modeling: Anscombe's quartet shows identical summary statistics can hide very different structure. (low confidence) — [source](https://en.wikipedia.org/wiki/Exploratory_data_analysis) ## How it works - Data analysis is the process of inspecting, cleansing, transforming and modeling data to discover useful information, inform conclusions and support decision-making; it spans many techniques under many names across business, science and social science. (high confidence) — [source](https://en.wikipedia.org/wiki/Data_analysis) - Tukey's five-number summary (min, quartiles, median, max) is preferred in EDA because median and quartiles are defined for all distributions and are more robust to skewed or heavy-tailed data than mean and standard deviation. (medium confidence) — [source](https://en.wikipedia.org/wiki/Exploratory_data_analysis) ## Comparisons and alternatives - Data mining is the subset of analysis that targets predictive rather than purely descriptive modeling and knowledge discovery; business intelligence is analysis that leans on aggregation of business information. (low confidence) — [source](https://en.wikipedia.org/wiki/Data_analysis) ## Facts and statements - Tukey's 1961 definition includes planning how data is gathered so analysis is easier, more precise or more accurate, not only analyzing data already in hand. (high confidence) — [source](https://en.wikipedia.org/wiki/Data_analysis) - Statistical data analysis divides into descriptive statistics, exploratory data analysis (discovering new features) and confirmatory data analysis (confirming or falsifying existing hypotheses). Predictive analytics and text analytics are further varieties. (medium confidence) — [source](https://en.wikipedia.org/wiki/Data_analysis) - Initial data analysis (IDA) checks assumptions needed for model fitting and hypothesis testing, handles missing values and applies variable transformations; EDA encompasses IDA. (medium confidence) — [source](https://en.wikipedia.org/wiki/Exploratory_data_analysis) - The analysis phases (data requirements, collection, processing, cleaning, exploration, modeling, communication) are iterative: feedback from later phases can send work back to earlier ones. (medium confidence) — [source](https://en.wikipedia.org/wiki/Data_analysis) - Do not generate and test a hypothesis on the same data: testing hypotheses suggested by the data introduces systematic bias, so hold out data or collect fresh data for confirmation. (medium confidence) — [source](https://en.wikipedia.org/wiki/Exploratory_data_analysis) - Common frameworks and tools for large-scale analysis include Hadoop (distributed batch processing), Spark (faster, more flexible cluster computing), Flink, Storm, RapidMiner (integrated text mining, prep, predictive analytics), Tableau (visualization/BI) and SAS. (low confidence) (verified as of 2026-10-18) — [source](https://en.wikipedia.org/wiki/Data_analysis) - Tukey's EDA drove the S language at Bell Labs, which inspired S-PLUS and R, and went together with robust and nonparametric statistics (jackknife, bootstrap resampling). (medium confidence) — [source](https://en.wikipedia.org/wiki/Exploratory_data_analysis)