Looks like no one added any tags here yet for you.
interpreting r²
____ of the variation in the ___ values can be explained by the line of regression
log-log
log y hat = a + b(logx)
MINI TAB finding r
r = sq root of r²
interpret scatter plot
strong positive linear relationship, negative, moderate, no relation
residual plot linearity
a pattern creates a problem with the linearity (curved residuals/outliers), you want random residuals above and below 0 line
extrapolation
never predict y-values higher than your highest data point or lower than your lowest data point
write line of regression
(predicted # of ____) = a + b(# of ___)
or y hat = a + bx, where y hat = predicted # of ___ and x = # of ___
slope
every time the # of x goes up by 1, the predicted # of y goes up by b(slope), on average
y - intercept
when the # of x is 0 the predicted # of y is a(y-intercept), on average
MAGIC: r²
r² is the proportion of variation in the y-values explained by the line of regression
finding a residual plot
y- y hat
Interpreting Se
Se is the average y - distance your points are away from the line of regression
Finding Se
Se = sq root of ssresid/n-2 (Find ss resid through sum(L3²)
transformations
transform data until linear than run linear regression (logs, log L1 and L2 into columns L3 and L4 then find linear regression)
MINITAB OUTPUT y intercept
constant, coefficient
MINI TAB slope
____, coefficient
MINI TAB r²
r-sq
MINI TAB Se
S
chance experiment definition
situation in which there is uncertainty about outcome
chance experiment examples
rolling a six sided die once, picking one card from a deck
Sample Space
set of all possible outcomes of a chance experiment
mutually exclusive/disjoing
two events are mutually exclusive or disjoint if they both can not happen at the same time
classical probability
if every outcome in the sample space of a chance experiment is equally likely then the probability of E or P(E) is P(E) = the # of outcomes including #/total # of outcomes in sample space (Works as long as outcome is equally likely)
Law of Large Numbers
As the # of repetitions of a chance experiment increases its relative frequency approaches the true probaility. If the # of trials is very large then: P(E)=The # of times E occurs/The total # of trials
probability rule
probability is always between 0 and 1
The NOT Rule
P(Not E) = 1 - P(E)
The OR Rule
P(E or F) = P(E) + P(F) - P (E And F)
If E and F are mutually exclusive in the OR rule..
P(E or F) = P(E) + P(F)
The AND rule
basis of sampling, P(A and B) = P(A) x P(B/A)
P(B/A)
The probability that B occurs GIVEN THAT A occurs: conditional probability (only way to string successive trials in a chance experiment, Ex: roll die twice