Help
Options
Didn't know it?
click below
click below
Knew it?
click below
click below
Don't Know
Remaining cards (0)
Know
retry
shuffle
restart
0:00
Embed Code - If you would like this activity on your web page, copy the script below and paste it into your web page.
Normal Size Small Size show me how
Normal Size Small Size show me how
PDAM #8
| Question | Answer |
|---|---|
| What is Logistic Regression used for (2) | 1, to give of the likelihood of output > odds 2, used when dependent variable is categorical |
| What is the shape of Logistic Regression (2) | 1, it is into logarithmic function transformed linear regression 2, it has kinda s shape (non-linear) |
| What is the formula of Logistic Regression | p = 1 / 1+e^(-t) where t = β0 + β1*X1 + β2*X2 |
| How to make Logistic Regression again linear | putting p = 1 / 1+e^(-t) into logit(p) = ln(p / 1-p) NECESSARY TO MAKE THE MODEL CORRECT |
| What is interpretation of p = 1 / 1+e^(-t) | the percentual likelihood of something happening |
| What is the goal while calculating logistic regression models | finding βs with the smallest LME (maximum-likelihood estimators) |
| How to calculate deviance of logistic regression (3) | 1, compute Log-Likelihood with LL =∑i [Yi*ln(pi) + (1-Yi)*ln(1-pi)] 2, Deviance = (-2) * LL 3, the smaller the better |
| How to tell if the Logistic Regression model is good | we compare the Deviance when using just the Average likelihood (Baseline) and the Deviance should drop signficantly |
| What are assumption and quality standarts of Logistic Regression (5) | 1, Dependent variable is dichotomous 2, outcomes are independents 3, outcomes are mutually exlusive 4, large samples min- 20-30 cases/predictor with min n=60 5, Multicollinearity can create confusion or misleading results |
| What are Evaluation Measures for Logistical Regression Models (4) | 1, Cox & Shell R^2, Nagelkerke R^2 2,Omnibus test of model coefficient 3, Hosmer and Lemeshow Test 4,Classification accuracy |
| How does A, Omnibus test of model coefficient (2) B,Hosmer and Lemeshow Test (2) work | A, Omnibus test of model coefficient >> H0: Proposed model not better than baseline (average) H1: Proposed model better than baseline B,Hosmer and Lemeshow Test >> H0: Model fits the date But very sensitive so DO NOT RELY ONLY ON THIS |
| How does Classification Accuracy works (2) | Hit Ratio = (Correct classification frequencies) / n more than 50% hit ratio is desired |
| How to compare competing models (2) | 1, with Cox & Shelly or Nagelkerke R^2 2, we need to use THE SAME DATA for that |
| How does VIF work | how much of variable influence on Y can be explained by others variables ("sumed" correlations with others) |
| What are goals for Factor Analysis and Principal Component Analysis (4) | 1, understanding the structure of a set of variables 2, constructing scale to measure a concept 3,reducing dataset to more manageable size 4,solving multicollinearity problems |
| What is the difference between PCA and FA | PCA - maximized total variance in the data FA - maximize commin variance in the data |
| What is the phylosifical idea behind Factor Analysis (2) | 1, finding "hidden" common factor that influence the variables 2, the r for variable group for one factor should be as high as possible and as low as possible for variables groups from other factors |
| What is the formula for a Variable in Factor Analysis | Variable = Loading x Common Factor x Unique Factor (error) |
| What is the formula for a Component in Principal Component Analysis | Component= Loading x Variables |
Created by:
semester1.IBWL