What impact does the exclusion of the metrics that represent minorities and language limited individuals have on the predictability of the CDC’s SVI, based on the 2018 data (CDC, 2018a; CDC, 2018b)?
Does the CDC’s SVI have key characteristics that impact the preclude potential exclusion without limiting the overall predictability of the SVI, based on the 2018 data (CDC, 2018a; CDC, 2018b)?
Data: • The data and data dictionaries are online. o Center for Disease Control and Prevention. (2018a). Social Vulnerability Index [data set]. https://svi.cdc.gov/Documents/Data/2018_SVI_Data/CSV/SVI2018_US.csv o Center for Disease Control and Prevention. (2018b). Social Vulnerability Index [code book]. https://svi.cdc.gov/Documents/Data/2018_SVI_Data/SVI2018Documentation.pdf o Note: Your raw data must be this report in its original form. Use the data dictionary to understand the data.
• Create a subset of the data. Consider the metrics that are used in creating the SVI. Use the data dictionary to identify the state variable field, along with the appropriate fields that represent the SVI metrics, considering the research questions. The SVI index’s variable name is RPL_THEMES, in column 99. o Socioeconomic
▪ Persons below the poverty estimate
▪ Civilian unemployed estimate
▪ Per capita income estimate
▪ Persons with no high school diploma o Household and composition disability features
▪ Ages 65 and older
▪ Ages 17 and under
▪ Persons with a disability, over the age of 5
▪ Single-parent households o Minority status and language limitations
▪ Persons with minority status
▪ Persons with no or minimal use of the English language o Housing types and transportation
▪ Multi-unit dwellings (10 or more units)
▪ Mobile homes
▪ Homes with more residents than a home is designed for
▪ Homes with no vehicle
▪ Group quarters or institutionalized quarters Note: Do not use the columns that are follow-on calculations of these columns. Variable names preceded with “E_” are actual measures, while “M_” represents the margin of error estimates. Do not include the margin of error estimates at this time. Considering the research questions, after subsetting and excluding the variable that houses the STATE field, there will be 13 columns with relevant information for analysis. The state field can be utilized in data exploration. Data Cleaning:
• Only clean the elements of the data that are inaccurately represented, such as missing values and data types.
• Do not remove missing values during cleaning. If missing values need to be removed for analysis method, do it during the preparation for analysis.
• When changing data values or types, ensure that you validate that the change occurred and the change is what was expected. Analyze:
• Conduct two types of analysis: exploratory data analysis and a random forest model. You will move through the sub-stages of Analyze two times; profile, prepare, and apply is required for each method of analysis in your program. It is consolidated in your research paper. Make sure that your analyses align with the research questions.**
• During exploratory data analysis, exclude visualizations in your final program or research paper that do not serve a purpose toward the objective of the research or represent meaningful information. Just remember, the relationships that do not exist in the data are sometimes as significant as the relationships that do exist.
o Define your plan for all analyses.
o Your plan for the random forest model will include splitting the data after seeding it so that training and testing evaluations can be conducted; additionally, you will exclude missing values from the model.
• Prepare o Carry out actions on the data that are necessary to prepare the data for the analyses.
• Apply (or Analyze) o When carrying out the random forest modeling, it may be imperative to understand how much time your analyses may take.
o Ensure that you analyze the results and understand what the different outcomes of the model represent. Provide interpretations of the overall model that is trained, how the trained model performs with the test data, and what features are most important for the explained variance. Results, Impact of the Results:
• Ensure that all analyses, visual or otherwise included in the final version of the program, include interpretations of what these analyses indicate.
• Make sure that you do not speculate! Use evidence to support any assertions that you make.
• You must also include recommendations for future analysis. An example might look something like this: o An opportunity for further research, based on gaps found in the random forest modeling, is to look at the ability to tune the parameters further, to improve the outcome.
o Additionally, an opportunity for future research is exploration modeling to determine what other variables, when eliminated, have little or no impact on the ability to predict the SVI based on the supporting characteristics in the data.
• You will base your recommendations on your findings in the analysis you conduct. Bonus challenge: Create a random forest model for each state you were assigned. Is there a difference in the models’ results? What does that result mean in terms of the data? Make sure that you do not speculate! Use evidence to support any assertions that you make. Required files to submit:
1) Research paper in APA 7 format; MS Word document file type
2) R Script; final version Good to know:
o The research paper will be written in a professional writing style, following APA 7 student paper format; you can use the student paper template.
o The document shall be 3-5 pages or at least 800 words. The page count does include the cover page or reference page. o Ensure that every reference in your reference list is also cited in the text. Do not forget to cite and reference the source of the data.
• When developing your research paper, you may modify the topic, problem, and research questions. However, the minimum requirements for the method of analysis cannot be altered.
• Ensure that you make the research yours and complete this assignment independently.
• There are several different versions of this assignment. If you complete a version of this assignment that is not available to you in Blackboard, you will violate your pledge.
• If you are concerned about how long the model is taking to run, you have a few options. o With the library GuessCompx, you can use the function CompEst() to evaluate the complexity and time it will take to process the function. Some things have to be done to the setup to use this methodology, though.
▪ Example: If the original model is called model <- train(outcome~.,data=d,subset=trainer,importance=T,method=”rf”) You would rewrite this as f <- function(d) train(outcome~.,data=d,subset=trainer,importance=F,method=”rf”) CompEst(d=d,f=f,random.sample=F)
o With the library doParallel, you can use the functions makePSOCKcluster() and registerDoParallel() you can run models in parallel, to improve performance.
▪ Prior to processing the model, use the following as an example, use help or Google to learn more. This will maximize the processor power.
cl <- makePSOCKcluster(4) # the value is the number of processor cores in your device registerDoParallel(cl) model <- train(outcome~.,data=d,subset=trainer,importance=T,method=”rf”,allowParallel=T) stopImplicitCluster() registerDoSEQ()