Datasets for Statistical Analysis: Regression

• Welcome
• Datasets from books
• Techniques index
  • time series
  • regression
  • generalized linear models
  • multivariate analysis
  • designed experiments
  • one- and two-sample data
• Applications index
• Reported statistics

Regression

• Simple linear regression
• Multiple regression
• Regression with factors
• Transformation
• Outliers
• Non-linear

Simple linear regression

• Artifical data (based on Pythagorus' theorem)
  Artificial data based on Pythagorus' Theorem, showing regression diagnostics can fail to indicate non-linear relationships
• AFL Football Department spending and results
  Spending of AFL clubs compared to performance
• Desorbtion of carbon monoxide
  A designed experiment with 22 observations; the variance could be non-constant
• Boiling point in the Alps
  A small, simple example of regression.
• Food prices
  A small, simple example of regression. There may be two outliers. A log-transform is probably needed to produce constant variance.
• Two assay experiments for a hormone
  A small, simple example of regression with non-constant variance.
• TV advertisements
  A small, simple example of regression, but with many issues to be considered before a good model is found.
• Alcohol and tobacco (not for release)
• Fish prices
  A small, simple example of regression with non-constant variance
• Shocking rats
  A small, simple example of regression. There is a suggestion of non-linearity.
• CEO salaries
  A small, simple example of regression. The regression, however, is not significant.
• Height and pulse rate
  The relationship is not very strong.
• Cotton yields
• Height and resting pulse
  The heights and resting pulses for a sample of hospital patients.
• Statures of brother and sister
  The heights of 11 pairs of brothers and sisters.
• Brisbane Lions crowds
  The number of members and average crowds sizes for the Brisbane Lions (formerly Brisbane Bears) Australian Rules Football club from 1987 (Brisbane's first year in the competition) to 2003 (but not till season's end).

Multiple regression

• Freeze drying
  A 3² factorial experiment on the effect of freeze-drying on biological material
• Body fat in females
  Body fat measurements of 20 healthy females, with three possible predictors
• Football punting
  While the two covariates are related to punting distance, fitting both produces marginal statistical significance.
• Chocolate bars
  
• Drug present in rat livers
  This dataset can be used to demonstrate influential points.
• Icecream sales
  This dataset can be used to demonstrate autocorrelated residuals in multiple regression.
• Naval hospital maintenance
  This dataset can be used to demonstrate collinearity in the predictors.
• House prices in Albuquerque
  This dataset contains missing values and possibly non-constant variance.
• Lunacy
  A knowledge of the geography is useful here.
• Life Expectancy
• Beef consumption in the USA
• The cost of finished steel
• Body dimensions
  Measurements on 21 body dimensions plus height, weight, age and gender for 507 physically active individuals.
• Manpower and workload for Navy Bachelor Office Quarters
  Manpower and workload for US Navy Bachelor Officers' Quarters

Regression with factors

• Freeze drying
  A 3² factorial experiment on the effect of freeze-drying on biological material
• Smoking and respiratory function
  The health smoking habits of 654 youths, aged 3 to 19 in the middle to late 1970s.
• Captopril and blood pressure
• Absorption of ions by potato
  The data can be used to test equality of slopes.
• Dimensions of jellyfish
  This dataset has two types of jellyfish, and a regression shows that they are indistinguishable on intercept and slope. But plot the data!
• Doses of ouabain to kill a cat
  A useful dataset to demonstrate weighted regression.
• Butterfat and dairy cattle
  Two factors
• Human age and fatness
  A small, simple example of regression with one factor and one quantitative covariate.
• Weight gain in rats
  A nice, small dataset for ANOVA with repeated observations.
• Tractor guidance systems
  Offsets from two tractor guidance systems taken over time, with swaths as factors.

Transformations

Multiple transforms possible

• Finnish fish catch data
  Measurements of 159 fish caught in the same lake in Finland

Reciprocal transformation

• Survival time of animals
• Lunacy
• Butterfat and dairy cattle

Log transformations

• Spider densitya>
  Density of certain spiders
• Demand for public transit services in rural areas
  Two small data sets for regression, with a number of potential predictors
• Elastomer viscosity
  A designed experiment measuring the viscosity of elastomers as a function of oil and filler
• Stress fractures in brittle material
  Fracture stress at five different levels of stress rates; requires a log transform of both variables.
• Two assay experiments for a hormone
• Fish prices
  
• Computer disc media
  Computer CD, DVD and DVD-RW prices in when purchased in packs. Useful for interpretting the regression parameters.
• Hooker's boiling point data
  The boiling point of water and air temperature in the Himalaya Mountains.
• Magazine revenue
  The revenue from magazine based on pages of advertising; non-constant variance and outliers

Outliers

• Cigarette data
  A small data set (25 observations on five variables) with an outlier, and two variables collinear
• TV advertisements
  A small, simple example of regression, but with many issues to be considered before a good model is found.
• Alcohol and tobacco (not for release)
• Food prices
• Lunacy
  A knowledge of the geography is useful here.

Non-linear regression

• Measurements from an electronic compass
  Observations from an electronic compass; ideally, the data follow a sine curve.

Also see:

• the generalized linear models datasets
• the designed experiments datasets