Cramer's V
Cramer's V is a normalized, directionless measure of association strength between categorical variables in a contingency table, ranging from 0 to 1.
Statistics Slam Dunk
Cramer's V is an effect-size measure of the strength of association between categorical variables in a contingency table.
- How it is used
- Measures the size of association between categorical variables after chi-square testing, using NBA rest and win-loss outcomes.
- How deep it goes
- Defines a 0–1 scale, demonstrates two R functions, and stops short of replacing significance testing.
- Examples
- Uses NBA days off, wins and losses, a contingency table, and questionr and rcompanion outputs of approximately 0.236.
- What you can do after
- Apply it via questionr or rcompanion to quantify association after constructing a contingency table.
Machine Learning for Tabular Data: XGBoost, Deep Learning, and AI
A normalized measure of association between categorical variables ranging from 0 to 1 and indicating association intensity without direction.
- How it is used
- Compares categorical features when correlation cannot represent their relationship reliably.
- How deep it goes
- Defines the normalized chi-square measure, interprets its range, and demonstrates calculation after discretization.
- Examples
- Uses MPG deciles, model year, a contingency table, and a demonstrated value of 0.855.
- What you can do after
- Apply it via SciPy and pandas to quantify association between categorical or discretized features.
Read how Machine Learning for Tabular Data: XGBoost, Deep Learning, and AI explains Cramer's V →
Related concepts
- Cohen's dcontrasts with · Cohen's d is a standardized effect-size measure expressing the difference between two group means in standard-deviation units, commonly using the pooled standard deviation.
- chi-square testcontrasts with · A chi-square test is a statistical hypothesis test that uses a chi-square-distributed statistic to evaluate hypotheses about categorical data, goodness of fit, or variance.
- effect sizekind of · An effect size is a quantitative measure of the magnitude of a difference, association, relationship, or other observed phenomenon, independent of statistical significance.
- chi-square statisticuses · A chi-square statistic measures the discrepancy between observed and expected frequencies, typically by summing squared differences divided by expected frequencies.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.