Knowra Exploratory data analysis Exploratory data analysis Exploratory data analysis uses graphical and numerical summaries to investigate a dataset, reveal patterns, and guide further analysis without relying primarily on formal confirmatory tests.
Data visualization : The graphical representation of data to reveal patterns, comparisons, and relationships. Plots make distributions and relationships visible in ways summary numbers alone may not.
Data : Recorded observations or measurements used as evidence for description, analysis, or decision-making. Exploration begins with observations whose structure and quality must be understood.
John Tukey : An American statistician known for influential work on data analysis, robust statistics, and computing. Tukey developed and popularized exploratory data analysis as a distinct approach.
Confirmatory data analysis : The evaluation of prespecified hypotheses or models using statistical procedures and explicit assumptions. It tests claims that exploratory inspection may suggest, rather than searching broadly for patterns.
Data cleaning : The process of detecting and correcting or documenting errors, inconsistencies, and unusable values in data. Exploratory summaries often reveal implausible values and inconsistencies that require cleaning.
Histogram : A graph that groups numerical observations into intervals and displays their frequencies as bars. Histograms show distribution shape, including skew, multiple peaks, and gaps.
Variable : A characteristic recorded for each unit in a dataset that can take different values. Variable type determines which summaries and plots are appropriate.
Statistical graphics : The use of graphical displays to summarize, analyze, and communicate statistical information. Tukey's approach built on a longer tradition of using graphs to investigate data.
Hypothesis testing : A statistical procedure that assesses evidence against a null hypothesis using a test statistic and reference distribution. Formal tests answer focused questions, while exploration often examines data before such questions are fixed.
Feature engineering : The creation or transformation of input variables to make data more useful for modeling. Exploring relationships and distributions can suggest useful transformations or derived variables.
Show all 25