Unit 3 Flashcards
What does r tell us
r= tells us strength and direction
r= how far away the points are from line of best fit
smaller r=
less correlation
how exact can the r be
1
high leverage points
tilt the line
far from the cluster in the x direction
influential points
have substantially large impact on slope, y-int, correlation
far from cluster in both x and y direction, doesnt fit pattern
formulas for find y-int and find slope
find y-int: a=ymean-bxmean
find slope: b=r(sy)/(sx)
choosing the best regression
-scatterplot: linear pattern
-r and r^2: close to r=1 and r^2=100%
-residual plot: no residual pattern
what does r^2 mean
the percent of variation
interpret r^2
ex- r^2= .974
___97.4%___ of the variation in __y__ is explained by the linear relationship with ____x_____
correlation does not equal causation
the r is .99 does not mean that x causes y its just showing an association
the x is called
and the y is called
explanatory
response
how to describe relationships
DUFS
Directions (pos/neg/none(flat))
Unusual features (outliers/ clusters)
Form (linear/nonlinear)
Strength (weak, moderate, strong)
Contex (relationship between __X__ and __y__)
scattered points on a scatterplot. linear or non linear?
linear bc nonlinear would have a curve
interpret the residual
the actual __y__ was _residual value was __above/below__ the predicted value for __x__
Dont forget units!
Explain the relationship displayed in the scatterplot
DUFS
D(direction)
U(unusual features)
F(linear, nonlinear)
S(stength)
explanatory and response on graph
x axis is explanatory and y axis is response
what is r^2
coefficient of determination
interpret coefficient of determination
___% of the variation in __y__ is explained by the linear relationship with __x__
how do you find the residual
actual-predicted (A-P)
point above the line
positive residual
point below the line
negative residual
find the correlation
ur looking for r
interpret slope
for each additional __x__, the predicted __y__ increases/decreases by __slope__
interpret y-int
when __x__=0 the predicted __y__ __contex__ is __y-int__
horizontal outlier
tilt the line
outlier that fit the y direction but not the x
vertical outlier
small impact
move line up or down
outlier that fit the x direction but not the y
