Continuous data is numeric data whose values can vary across a range, including fractional values. It is useful when measuring quantities such as age, games played, minutes played, win shares, test scores, or regular-season wins. Unlike categorical data, continuous values support questions about distributions, relationships, group differences, and prediction. section 1.2.1 section 5.7
Start by identifying the role of each variable. A histogram shows the distribution of one continuous variable. A correlation plot examines the association between two continuous variables, which may be positive, negative, or neutral. Boxplots can compare the distribution of a numeric outcome across categorical groups. section 1.2.1 section 14.5.1
Compare a correlation plot, faceted boxplots, a categorical bar chart, and a histogram to see how chart choice follows the type of data being analyzed. section 1.2.1
For two continuous variables, correlation summarizes their linear relationship with a coefficient from -1 to +1. Values near +1 indicate a positive relationship, values near -1 a negative relationship, and values near 0 little or no linear relationship. A strong correlation does not establish causation. section 2.5.4
Checking relationships among player measures
Select numeric variables such as Age, G, MP, WS, and WS48 before computing a correlation matrix in R. The resulting matrix lets you inspect every pair: games played and win shares, for example, show a strong positive correlation in the displayed output. A heat map can then color and label the coefficients for faster comparison. section 2.5.4
Continuous data also supplies the outcome for common statistical models. Multiple linear regression predicts a continuous target from two or more predictors; predictors are often continuous but may be categorical. For example, regular-season wins can be modeled as a value on a 0–82 scale, while a yes-or-no outcome calls for logistic rather than linear regression. section 5.7
When comparing two groups with numeric outcomes, t-tests assess whether an observed difference is statistically significant, while Cohen’s d measures the size of that difference. Comparing home and visiting teams’ foul-call or free-throw counts turns an NBA question about possible officiating bias into a testable analysis. section 7.0
Numeric variables can also be converted into categories when an analysis needs a categorical predictor. Standardized points scored or allowed can be divided into ranges, converted to factors, and plotted against standardized wins. This transformation is deliberate: numeric summaries return descriptive statistics, whereas factors return category counts. section 14.5.1
