Quiz 6 Flashcards
(24 cards)
Non-Linear regression
- fits data to a variety of possible curves
bacterial infection
- bacteria reproduced by fission (splitting)
- each original cell gives rise to two daughter cells
what does each variable mean for this equation?
Y=b+ex^m
y=population
b=parameter constant
e=etime
m=estimates b1
what uses 4 parameter logistic regression?
- biochemistry
- physiology
- population models
what is a 4 parameter logistic regression in physiology for?
- hormone levels and hemoglobin
what is a 4 parameter logistic regression in population growth for?
population growth
4 parameter logistic regression use what 4 measurements?
- Upper Asymptote
- Lower Asymptote
- Inflection point
- Slope at inflection point
On a 4 parameter logistic regression graph, where do you see the Upper Asymptote?
the top of the line
On a 4 parameter logistic regression graph, where do you see the lower Asymptote?
where the curve starts
On a 4 parameter logistic regression graph, where do you see the Inflection point?
where it changes from positive to negative
On a 4 parameter logistic regression graph, where do you see the slope at inflection point?
what is the slope at the inflection point?
Problems with 4 parameter logistic regression graph.
- Produces poor estimation
- Mis predicting
How many hospitalizations involve mechanical ventilators?
- as much as 3%
mortality is higher in what age group?
4-10x higher for patients over 70.
If we do not have normal distribution, and it is independent, what test will be used?
Mann-Whitney U Test
when do you use regression techniques?
- for different shapes
- regressions can also be used to predict variation in categorical responses
multivariate equation
y= m1(x1)+m2(x2)+m3(x3)+b
when you include more predictors for multivariate, what needs to happen to R^2?
it will need to increase
what does the post hoc test do?
tells us which predictors matter
more predictors means what?
model is better at predicting the noise in your data
real world dirty approach
include only significant predictors
real world good approach (AIC)
Akaike Information Criterion
- runs from -1 to 1
- 1 means did something wrong
1 means great
0 is not as efficient
when reporting new R^2, what value do you report?
use the adjusted R square column
what is R^2?
a measure of model fit
