Environment Setup<\/h2>\n
The following tutorial is carried out in RStudio (R version 4.0.2) on Windows 10. However, the syntax would remain the same irrespective of the Operating System used.<\/p>\n
Install and Load the necessary packages<\/h3>\n\ninstall.packages(\"dplyr\")\ninstall.packages(\"ggplot2\")\nlibrary(dplyr)\nlibrary(ggplot2)<\/code><\/pre>\n<\/div>\nLoad the Dataset<\/h3>\n
The dataset needs to be in the same directory as your R Working directory, otherwise, the below-mentioned command would output an error message.<\/p>\n
load(\"brfss2013.RData\")<\/code><\/p>\n.RData<\/code> is an extension of datasets particularly belonging to R. <\/p>\nTroubleshooting<\/h3>\n
Get your working directory using:<\/p>\n
getwd()<\/code><\/p>\nIn case your working directory is different from the location of your dataset, set your working directory using the following command. Alternatively, move the dataset to the above output location.<\/p>\n
setwd(dir)<\/code><\/p>\nWhere dir is a character string specifying your desired path. For more information on how to set your path, see here<\/a>.<\/p>\nAn easy alternative – GridDB<\/h2>\n
The above-mentioned dataset is nearly 600 MB. While this may not seem a big deal, uploading such an enormous amount of data on RCloud does not result in a fruitful outcome. Instead, the server takes a lot of time uploading this dataset due to which it eventually times out. This chunk of information is important because the tutorial we have carried out needs to be done using your local system. This does not leave room for scalability, especially if you\u00e2\u20ac\u2122re working in an organization which more often than not involves collaborating with multiple people. <\/p>\n
To solve this problem, we have a highly scalable and just the right solution for your data – GridDB<\/a>. GridDB is a highly scalable, reliable and relatively faster tool for your data storage. It also supports a number of programming languages including Java, C, Python, etc.<\/p>\nCheck out our quick tutorial on setting up Python-GridDB client here<\/a>. Alternatively, you can manually download<\/a> this open-source database especially optimized for IoT and Big Data.<\/p>\nFiltering and Visualization<\/h2>\n
Our foremost research focus for this tutorial is to visualize the distribution of a person\u00e2\u20ac\u2122s general health and whether or not it varies with a person\u00e2\u20ac\u2122s marital status. This research question would focus on two parameters from this dataset: General Health parameter (denoted by genhlth in the dataset) and the marital status (denoted by marital in the dataset).<\/p>\n
General Health of a person is subdivided into 5 main categories – Excellent, Very Good, Good, Fair, Poor and for some of the entries, this information might not be available (<NA><\/code>).<\/p>\nSimilarly, the possible values for the marital status are: Married, Divorced, Widowed, Separated, Never Married, or a member of an unmarried couple. Again, some of the participants might not have responded and those entries would have been filled with <NA><\/code>.<\/p>\nLet us first review what is the distribution of people under the general health category.<\/p>\n
\nbrfss2013 %>% \n group_by(genhlth) %>% \n summarise(count = n())<\/code><\/pre>\n<\/div>\nGives the following output:<\/p>\n
## # A tibble: 6 x 2\n## genhlth count\n## \n## 1 Excellent 85482\n## 2 Very good 159076\n## 3 Good 150555\n## 4 Fair 66726\n## 5 Poor 27951\n## 6 1985<\/pre>\nWe can see that 1985 people have not responded to the question at hand. Hence, we will be filtering out such entries.<\/p>\n
\nbrfss2013 %>% \n filter(genhlth != \"NA\") %>%\n group_by(genhlth) %>% \n summarise(count = n())<\/code><\/pre>\n<\/div>\nThe output should look something like this now.<\/p>\n
## # A tibble: 5 x 2\n## genhlth count\n## \n## 1 Excellent 85482\n## 2 Very good 159076\n## 3 Good 150555\n## 4 Fair 66726\n## 5 Poor 27951<\/pre>\nWe shall now have a look at the other parameter of interest: Marital Status. Again, we will be filtering out any missing entries.<\/p>\n
\nbrfss2013 %>% \n filter(marital!=\"NA\") %>%\n group_by(marital) %>% \n summarise(count = n())<\/code><\/pre>\n<\/div>\n## # A tibble: 6 x 2\n## marital count\n## \n## 1 Married 253329\n## 2 Divorced 70376\n## 3 Widowed 65745\n## 4 Separated 10662\n## 5 Never married 75070\n## 6 A member of an unmarried couple 13173<\/pre>\nNow, for visualizing every possible combination, we would need to aggregate this data based on these two parameters somehow. For this purpose, the dplyr library offers a function count which will be used to count people with excellent health and compare their marital status.<\/p>\n
\nagg = count(brfss2013, genhlth, marital) %>%\n filter(genhlth!=\"NA\", marital!=\"NA\")<\/code><\/pre>\n<\/div>\nLet\u00e2\u20ac\u2122s see what agg has stored.<\/p>\n
head(agg)\n-->\n## genhlth marital n\n## 1 Excellent Married 49682\n## 2 Excellent Divorced 10049\n## 3 Excellent Widowed 7007\n## 4 Excellent Separated 1221\n## 5 Excellent Never married 14419\n## 6 Excellent A member of an unmarried couple 2364<\/pre>\nNow that we have the count, let\u00e2\u20ac\u2122s see the proportion of married men\/women to their health.<\/p>\n
\nggplot(agg %>%\n filter(marital==\"Married\")) + geom_bar(aes(x = genhlth, y = n), stat = \"identity\")<\/code><\/pre>\n<\/div>\nThe above command gives the following output:<\/p>\n
![\"\"](\"https:\/\/griddb.net\/wp-content\/uploads\/2021\/01\/chart.png\")
<\/a><\/p>\nSimilarly, you could visualize this proportion for people with marital status other than married. Go ahead, plot some fancy plots with this simple code!<\/p>\n
Conclusion<\/h2>\n
In this tutorial, we analyzed the public BRFSS dataset<\/a> using the R programming language. We discussed some limitations of using a huge dataset on your local system and an easy, much faster alternative<\/a>.<\/p>\n
install.packages(\"dplyr\")\ninstall.packages(\"ggplot2\")\nlibrary(dplyr)\nlibrary(ggplot2)<\/code><\/pre>\n<\/div>\nLoad the Dataset<\/h3>\n
The dataset needs to be in the same directory as your R Working directory, otherwise, the below-mentioned command would output an error message.<\/p>\n
load(\"brfss2013.RData\")<\/code><\/p>\n.RData<\/code> is an extension of datasets particularly belonging to R. <\/p>\nTroubleshooting<\/h3>\n
Get your working directory using:<\/p>\n
getwd()<\/code><\/p>\nIn case your working directory is different from the location of your dataset, set your working directory using the following command. Alternatively, move the dataset to the above output location.<\/p>\n
setwd(dir)<\/code><\/p>\nWhere dir is a character string specifying your desired path. For more information on how to set your path, see here<\/a>.<\/p>\nAn easy alternative – GridDB<\/h2>\n
The above-mentioned dataset is nearly 600 MB. While this may not seem a big deal, uploading such an enormous amount of data on RCloud does not result in a fruitful outcome. Instead, the server takes a lot of time uploading this dataset due to which it eventually times out. This chunk of information is important because the tutorial we have carried out needs to be done using your local system. This does not leave room for scalability, especially if you\u00e2\u20ac\u2122re working in an organization which more often than not involves collaborating with multiple people. <\/p>\n
To solve this problem, we have a highly scalable and just the right solution for your data – GridDB<\/a>. GridDB is a highly scalable, reliable and relatively faster tool for your data storage. It also supports a number of programming languages including Java, C, Python, etc.<\/p>\nCheck out our quick tutorial on setting up Python-GridDB client here<\/a>. Alternatively, you can manually download<\/a> this open-source database especially optimized for IoT and Big Data.<\/p>\nFiltering and Visualization<\/h2>\n
Our foremost research focus for this tutorial is to visualize the distribution of a person\u00e2\u20ac\u2122s general health and whether or not it varies with a person\u00e2\u20ac\u2122s marital status. This research question would focus on two parameters from this dataset: General Health parameter (denoted by genhlth in the dataset) and the marital status (denoted by marital in the dataset).<\/p>\n
General Health of a person is subdivided into 5 main categories – Excellent, Very Good, Good, Fair, Poor and for some of the entries, this information might not be available (<NA><\/code>).<\/p>\nSimilarly, the possible values for the marital status are: Married, Divorced, Widowed, Separated, Never Married, or a member of an unmarried couple. Again, some of the participants might not have responded and those entries would have been filled with <NA><\/code>.<\/p>\nLet us first review what is the distribution of people under the general health category.<\/p>\n
\nbrfss2013 %>% \n group_by(genhlth) %>% \n summarise(count = n())<\/code><\/pre>\n<\/div>\nGives the following output:<\/p>\n
## # A tibble: 6 x 2\n## genhlth count\n## \n## 1 Excellent 85482\n## 2 Very good 159076\n## 3 Good 150555\n## 4 Fair 66726\n## 5 Poor 27951\n## 6 1985<\/pre>\nWe can see that 1985 people have not responded to the question at hand. Hence, we will be filtering out such entries.<\/p>\n
\nbrfss2013 %>% \n filter(genhlth != \"NA\") %>%\n group_by(genhlth) %>% \n summarise(count = n())<\/code><\/pre>\n<\/div>\nThe output should look something like this now.<\/p>\n
## # A tibble: 5 x 2\n## genhlth count\n## \n## 1 Excellent 85482\n## 2 Very good 159076\n## 3 Good 150555\n## 4 Fair 66726\n## 5 Poor 27951<\/pre>\nWe shall now have a look at the other parameter of interest: Marital Status. Again, we will be filtering out any missing entries.<\/p>\n
\nbrfss2013 %>% \n filter(marital!=\"NA\") %>%\n group_by(marital) %>% \n summarise(count = n())<\/code><\/pre>\n<\/div>\n## # A tibble: 6 x 2\n## marital count\n## \n## 1 Married 253329\n## 2 Divorced 70376\n## 3 Widowed 65745\n## 4 Separated 10662\n## 5 Never married 75070\n## 6 A member of an unmarried couple 13173<\/pre>\nNow, for visualizing every possible combination, we would need to aggregate this data based on these two parameters somehow. For this purpose, the dplyr library offers a function count which will be used to count people with excellent health and compare their marital status.<\/p>\n
\nagg = count(brfss2013, genhlth, marital) %>%\n filter(genhlth!=\"NA\", marital!=\"NA\")<\/code><\/pre>\n<\/div>\nLet\u00e2\u20ac\u2122s see what agg has stored.<\/p>\n
head(agg)\n-->\n## genhlth marital n\n## 1 Excellent Married 49682\n## 2 Excellent Divorced 10049\n## 3 Excellent Widowed 7007\n## 4 Excellent Separated 1221\n## 5 Excellent Never married 14419\n## 6 Excellent A member of an unmarried couple 2364<\/pre>\nNow that we have the count, let\u00e2\u20ac\u2122s see the proportion of married men\/women to their health.<\/p>\n
\nggplot(agg %>%\n filter(marital==\"Married\")) + geom_bar(aes(x = genhlth, y = n), stat = \"identity\")<\/code><\/pre>\n<\/div>\nThe above command gives the following output:<\/p>\n
<\/a><\/p>\nSimilarly, you could visualize this proportion for people with marital status other than married. Go ahead, plot some fancy plots with this simple code!<\/p>\n
Conclusion<\/h2>\n
In this tutorial, we analyzed the public BRFSS dataset<\/a> using the R programming language. We discussed some limitations of using a huge dataset on your local system and an easy, much faster alternative<\/a>.<\/p>\n