Regression Analysis Flashcards
Curve fitting
Describes the relationship between 2 variables
Standard curve or regression line = widely used
Standard curve axes
Concentration = x axis
Observed reading = y axis
Standard curve
Group of standards in increasing concentration
Record an analytical parameter
Estimate concentration of unknown w/ interpolation
Assumptions: linear regression
X axis values are essentially error free
Y axis values may have an error associated with them
Background interference
Weak signals at 0 concentration
Equipment reads something even if the substance isn’t present
Coefficient of Determination
R^2
Proportion of variation in the dependent variable that is predictable from the independent variable
Degree of correlation
R^2 drawbacks
Doesn’t tell you if the model is good at predicting the outcome
Doesn’t indicate if the model is adequate
RSME
Root mean square error
Low values = better fit, higher accuracy
Standard deviation of the residuals
How concentrated the data is around the line of best fit
SS Residual (Residual sum of squares)
Level of variation in the error term
Smaller value = better fit
Residual = observed - predicted
95% Confidence bands
More accurate y values near the center of the line
Ends have a higher variation, more error
F ratio
Variance between groups / variance within groups
F > reference → reject the null
Q test
Q = Next closest value - value / Range
Null hypothesis
No statistical difference between the two groups/variables
