DAA_06_Linear-models

Which are linear models

• anova

• correlation

• chi-square

• Kruskal-Wallis

Pearson Correlation ranges

• from -1 to 1

• from -∞ to ∞

• from 0 to 1

• from -1 to 0

A correlation value of 0 means

• There is no errors and therefore correlation is perfect

• There is a perfect positive correlation

• There is no correlation at all

• Correlation cannot be 0

A correlation value of -1 means

• There is no errors and therefore correlation is perfect

• There is a perfect positive correlation

• There is no correlation at all

• There is a perfect negative correlation

R² coefficient of determination ranges

• from -1 to 1

• from -∞ to ∞

• from 0 to 1

• from -1 to 0

A R² value of 0 means

• The sum of errors is 0 and therefore the model is perfect

• There is a negative relationship

• There is no variance explained by the model

• R² cannot be 0

The p value of a linear model means

• The sum of all errors

• The probability for the slope of the line to be 0

• The probability for having an effect of x on y

• The likelihood of the model.

A linear model with p = 0.000001 means

• There is an effect of x on y

• There is no effect of x on y

• There is an effect of y on x

• The model is well fitted

Assumptions of a linear model

• Random, independent and representative sampling

• Normal distribution of dependent variable

• Normal distribution of residuals

• Homoscedasticity

• Normal distribution of independent variable