Quant II, Summer 2014

[cwyoo]

Please reply to this post to post all materials, such as project proposal, final paper, and their related questions/answers.

[Tharaa]

Hello all,
The attachment shows an example of applying linear regression analysis on body fat data, and its interpretation.

[Tharaa]

Here is the results of binary logistic regression using SPSS for same body fat data.

[cwyoo]

Hello all,
The attachment shows an example of applying linear regression analysis on body fat data, and its interpretation.

Great work! Can any of you comment on Pearson's correlation results with the linear regression results?

[cwyoo]

I attached 3 files:
1- Class proposal.
2- Bodyfat results-Linear Regression results.
3- PAH results-Linear Regression results.

Banjo's bayesian networks are still in process.

Can you post the class related materials under Manuscripts & Documentation > Class Projects > QuantII? There, please discuss about the results in bodyfat linear regression, e.g., what does the R squre suggests? Are results supporting linear relationships among the input variables and outcome? Did this dataset meet the assumptions of linear regression?, etc.

Hamza, these are comments about your class project for quant II (Tharaa, I ask you to discuss with Hamza when you are writing the Method section):
- Bayesian network analysis is for your previous class project. You may move it to there. Please submit the previous class project report separately.
- Write more about linear and logistic regressions in the Method section. Make subsections named "Linear regression" and "Logistic regression" and describe assumptions, how the regression is implemented (use math equations, if needed), and how the results are interpreted.

[cwyoo]

Here is the results of binary logistic regression using SPSS for same body fat data.

Great job!

Things to try:
- Is there any other option to enter the independent variables into the model? There is a method called forward selection and backward elimination. I believe SPSS implements this method. Can any of you find out how to use this method in analyzing the body fat data?
- Comment on the results.
- Based on the "Classification Table" on Step 1, what is the specificity and sensitivity of this model?

[Tharaa]

Here is my interpretation for logistic regression analysis:
•The Cox & Snell R Square and Nagelkerke R Square values are sometimes called (pseudo R2) values, and explain the variation of body fat.
•According to summery model table, the variation ranges from 52.5% to 70%.
•In classification table, the cut value = 0.5 that means if the case probability above 0.5 then it is above body fat mean (1), and if the case probability below 0.5 then it is below body fat mean (0).
•Percentage accuracy in classification (PAC)= 86.5%
•Sensitivity= 85.3%
•Specificity= 87.8%
•The positive predictive value= 88%
•The negative predictive value= 85%
•In Variables in the Equation table, the sig. column shows the significance of each variable. For example, age do not add significance to the model as (P= 0.226).
•In Variables in the Equation table, the Exp(B) column shows the odds ratio. For example, the increase in abdomen measures increases the likelihood of high body fat. In contrast, the increases in hip and biceps measures reduce the likelihood of high body fat.

[Hamza]

In attachment you will find my results for body fat linear regression.

My interpretation:
• Since R2 = 0.749, the model explains that about 75% of the variability in the response of bodyfat.
• Based on the p-value of the F test in ANOVA table, we can conclude that, the results support the linear relationship between bodyfat and predictors. Also, normal P-P plot represent linear relationship for dependent variable which is bodyfat.
• Pearson Correlation represented that there are strong relationship between bodyfat and many variables including weight, chest, abdomen, hip, thigh, and knee. However, there were a weak relationship between bodyfat and many variables including age, neck, ankle, biceps, forearm, and wrist. Also, there were a negative relationship between bodyfat and height, this means when one variable increase in value, the other will decrease in value.
• For Unstandardized Coefficients table, we can conclude based in p-value that there are a significant association between bodyfat and many variables including age, neck, abdomen, foreaem, and wrist; which all p-value <0.05.

[Hamza]

Here is updated proposal for Quant Analysis II class.

[Hamza]

This is an updated version of previous class report about PAH as requested. Results will be added.