Logistic Regression  

 

Assigned Reading

  • Session overview YouTube►
  • Read Chapter 12 in Statistical Analysis of Electronic Health Records by Farrokh Alemi, 2020
  • Calculate McFadden R-squared ChatGPT►
  • Replacing logistic regression with ordinary regression  Slides► YouTube► Video►

Assignment

Assignments should be submitted in Blackboard.  Include a summary page.  In the summary page, write statements comparing your work to answers given or videos.  For example, "I got the same answers as the Teach One video for question 1." Or you can write: "There was no answer sheet available for question 2."  We prefer that assignments are done in R.

Question 1: Use the following corpus of training data.  Classify if the target sentence is a complaint. The corpus is organized as in the following table. The comment ID shows the comment in the training data.  In the following table, 6 comments in the training set are displayed. The columns on the right of the table show where in the training comment the words from the target comment appears.  For example, in the training comment 57685 the word "patient" in the target comment is the third word in the training comment.

comment
Id		 Type
Id		 Classification
True = Complaint
False = Praise		 loves patients tell about not money
57685 1 TRUE 0 3 0 2 0 0
57688 1 TRUE 0 0 1 0 0 0
57703 1 TRUE 0 0 0 0 0 9
57704 1 TRUE 0 0 0 3 0 0
57711 1 FALSE 0 8 0 0 0 0
57712 1 TRUE 0 0 0 0 2 0

In the following, calculate predicted value of a logistic regression using the following formula:

  1. Regress the classification labels in the training set on the words, pair of consecutive words, and triplets of consecutive words in the target sentence: "He loves his patients and I can tell it's about us and not the money."  Use the predicted probability of complaint to classify the target sentence.  Values above 0.5 should be classified as complaints. 
  2. Regress the classification labels in the training set on the words, pair of words, triplet of consecutive words in the target sentence "However, I am not happy with rhinoplasty revision results." Use the predicted probability of complaint to classify the target sentence.  Values above 0.5 should be classified as complaints.
  3. Repeat the analysis but this time include all of the complaints and 50% random sample of praises in the training data set. How did the sampling procedure affect the McFadden R-square

Resources:

  • Labeled training data set for: "He loves his patients and I can tell it's about us and not the money." Download►
  • Labeled training data set for: "However, I am not happy with rhinoplasty revision results." Download►
  • How to predict response variable in logistic regression ChatGPT►
  • How to drop variables that are perfectly correlated? R Code►
  • Full Corpus (needed for analysis of other target comments) Download► Preprocessing ChatGPT►
  • Vladimir Cardenas's Answer► R Code►
  • Regina Reyes's Teach One on "However, I am happy with rhinoplasty revision results." Slides► YouTube►
  • Sravya's Teach One on "He loves his patients and I can tell it's about us and not the money." Slides► YouTube►

Question 2: Regress survival in next 6 months on disabilities of the patients, age of patients, gender of patients and whether they  participated in the medical foster home program. MFH is an intervention for nursing home patients.  In this program, nursing home patients are diverted to a community home and health care services are delivered within the community home.  The resident eats with the family and relies on the family members for socialization, food and comfort.  It is called "foster" home because the family previously living in the community home is supposed to act like the resident's family. Enrollment in MFH is indicated by a variable MFH=1. 

Survival is reported in two variables.  One variable indicates survival in 6 months.  Another reports days known to survive, if the patient has died and otherwise null.  Thus a null value in this latter variable indicates the patient did not die.  

The functional disabilities are probabilities that the patient has the disability.  These probabilities are generated from the CCS diagnoses and demographics of the person. Use long term disabilities. These are the disabilities with suffix 365.  If the disability is higher than 0.5, then assume the person is disabled. 

  1. Clean the data.  Convert the disabilities to binary variables.  Convert the age to decades
  2. Create a regression model to explain the relationship among the variables and survival. 
  3. List the top 4 predictors of survival (list these predictors using English language and not coded data). 
  4. Describe, in English, if the MFH program contributes to survival.  Provide the evidence for your claim.

Resources:

Question 3: The following data provide the length of stay of patients seen by Dr. Smith (Variable Dr Smith=1) and his peer group (variable Dr. Smith = 0).  Does Dr. Smith see a different set of patients than his peer group?  In particular, what is the probability of patients being seen by Dr. Smith.  Regress the choice of provider on the 9 diagnoses provided. 

Resources:

Question 4:  In a nursing home, data were collected on residents' survival and disabilities.  The data are listed in the following order: ID, age, gender (M for male, F for Female), number of assessments completed on the person, number of days followed, days since first assessment, days to last assessment, unable to eat, unable to transfer, unable to groom, unable to toilet, unable to bathe, unable to walk, unable to dress, unable to bowel, unable to urine, dead (1) or alive (0), and assessment number.  Predict from the patient's assessments (i.e. their age and current disabilities at time of assessment) if the patient is likely to die. Here are the steps in this analysis: 

  1. Read the data, making sure all entries are numbers. 
  2. Calculate age at each assessment not just at first assessment. 
  3. Clean the data, removing impossible situations (remove cases with date of assessment after death). 
  4. Remove irrelevant cases (all cases that have only one assessment)
  5. Organize age at current admission into a binary variable above or below the average age at current assessment.
  6. Estimate missing values
  7. Regress death in 6-months on various current disabilities, age, gender, and pairwise interactions of these variables. 

Resources:

Question 5: Repeat question 4 but now predict 6 month likelihood of first occurrence of walking disorders instead of death.  In this analysis, exclude all assessments that occur after walking disability has occurred. 

Question 6:  In a nursing home, data were collected on residents' survival and disabilities.  The data are listed in the following order: ID, age, gender (M for male, F for Female), number of assessments completed on the person, number of days followed, days since first assessment, days to last assessment, unable to eat, unable to transfer, unable to groom, unable to toilet, unable to bathe, unable to walk, unable to dress, unable to bowel, unable to urine, dead (1) or alive (0), and assessment number.  Predict from the patient's assessments (i.e., their age and disabilities at time of assessment) if the patient is likely to die and should be admitted to the hospice program. 

More

For additional information (not part of the required reading), please see the following links:

  1. Regression using R Read►
  2. Statistical learning with R Read►
  3. Open introduction to statistics Read►

This page is part of the HAP 819 course on Advanced Statistics by Farrokh Alemi PhD Home► Email►