R/analyzeRq3.R
In DataScienceSalon/chronic: Clinical Depression: Prevalence, SocioDemographic, Chronic Disease Comorbidity, and Costs
#==============================================================================#
# analyzeRq3.R #
#==============================================================================#
#'
#' analyzeRq3
#'
#' \code{analyzeRq3(brfss)}{Performs analyses for research question #3,
#' To what degree does depression with co-occurring chronic illness affect
#' productivity, vis-a-vis that of diagnoses of depression or chronic illness
#' separately?}
#'
#' This function analyzes the effects of depression and chronic illness on the
#' response variable, the number of sick days reported in the 30 days
#' preceding the survey. This analysis is conducted in the following five
#' sections.
#'
#' @section Acquire Data:
#' This section extracts the number of sick days and the sick days range
#' variable, as well as the depression and chronic illnesses indicator
#' variables.
#'
#' @section Create Marginal and Joint Contingency Tables:
#' This section renders the tables that illuminate the marginal and joint
#' effects of depression and chronic illness on the number of sick days
#' reported. Descriptive statistics of sick days are also reported
#' for both depression and chronic illness, as well as their joint
#' effects on the number of sick days reported.
#'
#' @section Plots:
#' In this section, frequency and proportion bar plots, box plots, violin
#' plots and histograms are rendered to illuminate the distribution of
#' sick days vis-a-vis chronic illness and depression diagnoses.
#'
#' @section Statistical Tests:
#' In this section, the non-parametric Wilcoxin signed-rank and the
#' the Kolmogorov–Smirnov tests were administered to ascertain the
#' effect of depression and chronic illness on the number of sick days
#' reported. The Wilcox test compared the mean number of sick days
#' reported for both depression and chronic disease diagnoses for
#' statistically significant differences. The Komogorov-Smirnov tests
#' were conducted pairwise to reveal relative effects of depression
#' and chronic illness on sick days reported.
#'
#' @param brfss Data frame of preprocessed 2013 Behavioral Risk Factor
#' Surveillance System (BRFSS) telephone survey.
#'
#' @return analysis List containing frequency and proportion data, plots,
#' and statistical tests. The variables are as follows:
#' \itemize{
#' \itemize{Data Tables}{
#' \item{depressionFreqTbl}{Table of frequencies of depression diagnoses}
#' \item{depressionPropTbl}{Table of proportions of depression diagnoses}
#' \item{chronicFreqTbl}{Table of frequencies of chronic disease diagnoses}
#' \item{chronicPropTbl}{Table of proportions of chronic disease diagnoses}
#' \item{interactionFreqTbl}{Table of frequencies of the interaction between chronic illness and depression diagnoses}
#' \item{interactionPropTbl}{Table of proportions of the interaction between chronic illness and depression diagnoses}
#' }
#' \itemize{Data Frames}{
#' \item{depressionData}{Depression frequency data in data frame format}
#' \item{chronicData}{Chronic disease frequency data in data frame format}
#' \item{interactionData}{Interaction between depression and chronic illness frequency data in data frame format}
#' }
#' \itemize{Statistics}{
#' \item{depression}{Descriptive statistics for sick days vis-a-vis depression diagnoses}
#' \item{chronic}{Descriptive statistics for sick days vis-a-vis chronic illness diagnoses}
#' \item{interaction}{Descriptive statistics for sick days vis-a-vis the depression / chronic illness interaction variable}
#' \item{allChronic}{Descriptive statistics for sick days vis-a-vis a diagnoses of any chronic illness}
#' }
#' \itemize{Plots}{
#' \item{depressionFreqBar}{The depression frequency bar plot}
#' \item{depressionPropBar}{The depression proportion bar plot}
#' \item{depressionHist1}{A histogram of the distribution of sick days vis-a-vis a diagnosis of depression}
#' \item{depressionHist2}{A histogram of the distribution of one or more sick days reported vis-a-vis a diagnosis of depression}
#' \item{depressionViolin}{A violin plot of the distribution of sick days vis-a-vis a diagnosis of depression}
#' \item{depressionBox}{A box plot of the distribution of sick days vis-a-vis a diagnosis of depression}
#' \item{chronicFreqBar}{The chronic illness diagnosis frequency bar plot}
#' \item{chronicPropBar}{The chronic illness diagnosis proportion bar plot}
#' \item{chronicHist1}{A histogram of the distribution of sick days vis-a-vis a diagnosis of chronic illness}
#' \item{chronicHist2}{A histogram of the distribution of one or more sick days reported vis-a-vis a diagnosis of chronic illness}
#' \item{chronicViolin}{A violin plot of the distribution of sick days vis-a-vis a diagnosis of chronic illness}
#' \item{chronicBox}{A box plot of the distribution of sick days vis-a-vis a diagnosis of chronic illness}
#' \item{allChronic}{A box plot of the distribution of sick days vis-a-vis a diagnosis of of any chronic illness}
#' \item{interactionFreqBar}{The depression / chronic illness interaction variable frequency bar plot of sick }
#' \item{interactionPropBar}{The depression / chronic illness interaction variable proportion bar plot}
#' \item{interactionHist1}{A histogram of the distribution of sick days vis-a-vis the depression / chronic illness interaction variable}
#' \item{interactionHist2}{A histogram of the distribution of one or more sick days reported vis-a-vis the depression / chronic illness interaction variable}
#' \item{interactionViolin}{A violin plot of the distribution of sick days vis-a-vis the depression / chronic illness interaction variable}
#' \item{interactionBox}{A box plot of the distribution of sick days vis-a-vis the depression / chronic illness interaction variable}
#' \item{allinteraction}{A box plot of the distribution of sick days vis-a-vis the depression / chronic illness interaction variable}
#' }
#' \itemize{tests}{
#' \item{depressionTest}{The Wilcoxin ranked-sums test of equal means with and without depression}
#' \item{depressionEffect}{The Kolmogorov-Smirnov test of the effect of depression on sick days reported}
#' \item{chronicTest}{The Wilcoxin ranked-sums test of equal means with and without a diagnosis of chronic illness}
#' \item{chronicEffect}{The Kolmogorov-Smirnov test of the effect of chronic illness on sick days reported}
#' \item{interactionTest}{The Wilcoxin ranked-sums test of equal means vis-a-vis the depression / chronic illness interaction variable}
#' \item{interactionModel}{The linear model that relates sick days to the values of the interaction variable}
#' \item{pairwise}{The pairwise Kolmogorov-Smirnov test of the effect of depression and chronic illness on sick days reported}
#' }
#' }
#'
#'
#' @author John James, \email{jjames@@datasciencesalon.org}
#' @export
analyzeRq3 <- function(brfss) {
#---------------------------------------------------------------------------#
# Get Data #
#---------------------------------------------------------------------------#
# Get complete cases for interaction analysis and create interaction variable
interactionData <- brfss %>%
filter(!is.na(SickDays) & !is.na(Depression) & !is.na(Chronic) & SickDays < 32) %>%
select(SickDays, Depression, Chronic, SickDaysInd)
interactionData$DepressionChronic <-
interaction(interactionData$Depression, interactionData$Chronic)
interactionDataPhd <- interactionData %>% filter(SickDays > 0)
# Get depression data
depressionData <- brfss %>% filter(!is.na(SickDays) &
!is.na(Depression) & SickDays < 32) %>%
select(SickDays, Depression, SickDaysInd)
depressionDataPhd <- depressionData %>% filter(SickDays > 0)
# Get chronic data
chronicData <- brfss %>% filter(!is.na(SickDays) &
!is.na(Chronic) & SickDays < 32) %>%
select(SickDays, Chronic, SickDaysInd)
chronicDataPhd <- chronicData %>% filter(SickDays > 0)
# Get data for all chronic disease, including depression
allChronicData <- brfss %>% filter(!is.na(SickDays) & Depression == 'No' &
SickDays < 32 &
!is.na('Heart Attack') & !is.na(Chronic) &
Chronic == 'Yes' &
!is.na('Angina Or Coronary Heart Disease') &
!is.na(Stroke) & !is.na(Asthma) &
!is.na('Skin Cancer') & !is.na('Cancer (Other)') &
!is.na(COPD) & !is.na(Arthritis) &
!is.na('Kidney Disease') & !is.na(Diabetes)) %>%
select(SickDays, Depression, SickDaysInd, 'Heart Attack',
'Angina Or Coronary Heart Disease', Stroke, Asthma, 'Skin Cancer',
'Cancer (Other)', COPD, Arthritis, Diabetes, 'Kidney Disease')
#---------------------------------------------------------------------------#
# Create Depression Marginal Contingency Tables #
#---------------------------------------------------------------------------#
# Create Tables for Analysis of Effect of Depression on Productivity #
#---------------------------------------------------------------------------#
# Create 2-Way Marginal Contingency Tables on Depression and Sick Days
depressionTbl <- with(depressionData, table(Depression, SickDaysInd))
depressionFreqTbl <- ftable(depressionTbl)
depressionFreqDf <- as.data.frame(depressionFreqTbl)
depressionPropTbl <- ftable(prop.table(depressionTbl, 1))
depressionPropDf <- as.data.frame(depressionPropTbl)
# Format 2-Way Depression Marginal Frequency Contingency Tables for plotting
depressionFreqDf <- depressionFreqDf %>% arrange(Depression,
desc(SickDaysInd)) %>%
group_by(Depression) %>%
mutate(cumFreq = cumsum(Freq), pos = cumFreq - 0.5 * Freq)
# Format 2-Way Depression Marginal Proportion Contingency Tables for plotting
depressionPropDf <- depressionPropDf %>% arrange(Depression,
desc(SickDaysInd)) %>%
group_by(Depression) %>%
mutate(pct = round(Freq * 100, 0), cumPct = cumsum(pct),
pos = cumPct - 0.5 * pct)
#---------------------------------------------------------------------------#
# Create Chronic Illness Marginal Contingency Tables #
#---------------------------------------------------------------------------#
# Create Tables for Analysis of Effect of Chronic Illness on Productivity #
#---------------------------------------------------------------------------#
# Create 2-Way Marginal Contingency Tables on Chronic Ilness and Sick Days
chronicTbl <- with(chronicData, table(Chronic, SickDaysInd))
chronicFreqTbl <- ftable(chronicTbl)
chronicFreqDf <- as.data.frame(chronicFreqTbl)
chronicPropTbl <- ftable(prop.table(chronicTbl, 1))
chronicPropDf <- as.data.frame(chronicPropTbl)
# Format 2-Way Chronic Marginal Frequency Contingency Tables for plotting
chronicFreqDf <- chronicFreqDf %>% arrange(Chronic,
desc(SickDaysInd)) %>%
group_by(Chronic) %>%
mutate(cumFreq = cumsum(Freq), pos = cumFreq - 0.5 * Freq)
# Format 2-Way Chronic Marginal Proportional Contingency Tables for plotting
chronicPropDf <- chronicPropDf %>% arrange(Chronic,
desc(SickDaysInd)) %>%
group_by(Chronic) %>%
mutate(pct = round(Freq * 100, 0), cumPct = cumsum(pct),
pos = cumPct - 0.5 * pct)
#---------------------------------------------------------------------------#
# Create Depression and Chronic Illness Contingency Tables #
#---------------------------------------------------------------------------#
# Create the 2X2 Depression and Chronic Illnesses Interaction Tables #
#---------------------------------------------------------------------------#
# Create Joint Contingency Tables on Depression and Chronic Illness
interactionTbl <- with(interactionData, table(Depression, Chronic, SickDaysInd))
interactionFreqTbl <- ftable(interactionTbl)
interactionFreqDf <- as.data.frame(interactionFreqTbl)
interactionPropTbl <- ftable(prop.table(interactionTbl, 1))
interactionPropDf <- as.data.frame(interactionPropTbl)
# Format 2-Way Depression Marginal Frequency Contingency Tables for plotting
interactionFreqDf <- interactionFreqDf %>% arrange(Depression, Chronic,
desc(SickDaysInd)) %>%
group_by(Depression) %>%
mutate(cumFreq = cumsum(Freq), pos = cumFreq - 0.5 * Freq)
# Format 2-Way Depression Marginal Proportion Contingency Tables for plotting
interactionPropDf <- interactionPropDf %>% arrange(Depression, Chronic,
desc(SickDaysInd)) %>%
group_by(Depression) %>%
mutate(pct = round(Freq * 100, 0), cumPct = cumsum(pct),
pos = cumPct - 0.5 * pct)
#---------------------------------------------------------------------------#
# Get Summary Statistics #
#---------------------------------------------------------------------------#
# Get mean of restricted activity days by depression
depressionMeans <- aggregate(SickDays ~ Depression, interactionData, mean)
depressionMeansPhd <- aggregate(SickDays ~ Depression, interactionDataPhd, mean)
# Get mean of restricted activity days by chronic
chronicMeans <- aggregate(SickDays ~ Chronic, interactionData, mean)
chronicMeansPhd <- aggregate(SickDays ~ Chronic, interactionDataPhd, mean)
# Get mean of restricted activity days by depression and chronic illness
interactionMeans <- aggregate(SickDays ~ Depression + Chronic, interactionData, mean)
interactionMeansPhd <- aggregate(SickDays ~ Depression + Chronic, interactionDataPhd, mean)
#---------------------------------------------------------------------------#
# Summary stats function
getStats <- function(theData, varName, eVar, value) {
stats <- psych::describe(theData)
stats <- data.frame(stats, row.names = NULL)
ci <- pastecs::stat.desc(theData)[11,1]
ds <- data.frame(Variable = varName,
Value = value,
N = stats$n,
Min = stats$min,
Lower = quantile(theData, .25, na.rm = TRUE),
Median = stats$median,
Mode = getMode(as.vector(theData$SickDays)),
Mean = round(stats$mean, 2),
CI = round(ci, 3),
Upper = quantile(theData, .75, na.rm = TRUE),
Max = stats$max,
Range = stats$range,
Total = sum(theData, na.rm = TRUE),
SD = round(stats$sd, 2),
SE = round(stats$se, 3),
Skew = round(stats$skew, 2),
Kurtosis = round(stats$kurtosis, 2))
names(ds)[names(ds) == 'Value'] <- eVar
return(ds)
}
#---------------------------------------------------------------------------#
# Get summary stats of sick days by depression, chronic illness, and interaction
values <- list('Yes', 'No')
depressionStats <-
data.table::rbindlist(lapply(seq_along(values), function(x) {
getStats(depressionData %>% filter(Depression == values[[x]])
%>% select(SickDays), 'SickDays', 'Depression', values[[x]])
}))
chronicStats <-
data.table::rbindlist(lapply(seq_along(values), function(x) {
getStats(chronicData %>% filter(Chronic == values[[x]])
%>% select(SickDays), 'SickDays', 'Chronic', values[[x]])
}))
# Chronic Illness Stats
depStats <- getStats(depressionData %>% filter(depressionData[,2] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Depression')
haStats <- getStats(allChronicData %>% filter(allChronicData[,4] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Heart Attack')
anginaStats <- getStats(allChronicData %>% filter(allChronicData[,5] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Angina or CHD')
strokeStats <- getStats(allChronicData %>% filter(allChronicData[,6] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Stroke')
asthmaStats <- getStats(allChronicData %>% filter(allChronicData[,7] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Asthma')
skinStats <- getStats(allChronicData %>% filter(allChronicData[,8] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Skin Cancer')
cancerStats <- getStats(allChronicData %>% filter(allChronicData[,9] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Cancer (Other)')
copdStats <- getStats(allChronicData %>% filter(allChronicData[,10] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'COPD')
arthritisStats <- getStats(allChronicData %>% filter(allChronicData[,11] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Arthritis')
diabetesStats <- getStats(allChronicData %>% filter(allChronicData[,12] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Diabetes')
kidneyStats <- getStats(allChronicData %>% filter(allChronicData[,13] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Kidney Disease')
allChronicStats <- rbind(depStats, haStats, anginaStats, strokeStats, asthmaStats, skinStats,
cancerStats, copdStats, arthritisStats, diabetesStats,
kidneyStats)
allChronicStats <- allChronicStats[order(-allChronicStats$Mean),]
values <- list('Yes.Yes', 'No.Yes', 'Yes.No', 'No.No')
interactionStats <-
data.table::rbindlist(lapply(seq_along(values), function(x) {
getStats(interactionData %>% filter(DepressionChronic == values[[x]])
%>% select(SickDays), 'SickDays', 'DepressionChronic', values[[x]])
}))
#---------------------------------------------------------------------------#
# Create Depression Plots #
#---------------------------------------------------------------------------#
# Create Marginal Frequency Bar Plot of Sick Days Indicator by Depression
depressionFreqBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Depression, y = Freq, fill = SickDaysInd),
data = depressionFreqDf, stat = 'identity') +
ggplot2::geom_text(data = depressionFreqDf, ggplot2::aes(x = Depression, y = pos,
label = Freq),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens',
name = "Sick Days") +
ggplot2::labs(title = "Frequency Distribution",
x = "Diagnosis of Depression",
y = 'Number Reporting Sick Days')
#---------------------------------------------------------------------------#
# Create Marginal Proportion Bar Plot of Sick Days Indicator by Depression
depressionPropBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Depression, y = pct, fill = SickDaysInd),
data = depressionPropDf, stat = 'identity') +
ggplot2::geom_text(data = depressionPropDf, ggplot2::aes(x = Depression, y = pos,
label = paste0(pct,"%")),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens', name = "Sick Days") +
ggplot2::labs(title = "Proportional Distribution",
x = "Diagnosis of Depression",
y = 'Percentage Reporting Sick Days')
#---------------------------------------------------------------------------#
# Create violin of sick days grouped by diagnosis of depression
depressionViolin <- ggplot2::ggplot(depressionData,
ggplot2::aes(x = Depression, y = SickDays, fill = Depression)) +
ggplot2::geom_violin(draw_quantiles = c(0.25, 0.5, 0.75)) +
ggplot2::labs(title = 'Violin Plot',fill = 'Depression') +
ggplot2::ylab('Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Create box of sick days grouped by diagnosis of depression
depressionBox <- ggplot2::ggplot(depressionData,
ggplot2::aes(x = Depression, y = SickDays, fill = Depression)) +
ggplot2::geom_boxplot() +
ggplot2::labs(title = 'Box Plot', fill = 'Depression') +
ggplot2::ylab('Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Plot a histogram of restricted activity days by depression diagnosis
plotDepressionHist <- function(plotData, means) {
# Create histogram of sick days grouped a diagnosis of depression
hist <- ggplot2::ggplot(plotData,
ggplot2::aes(x=SickDays, color = Depression, fill = Depression)) +
ggplot2::geom_histogram(ggplot2::aes(y=..density..), position = "identity", alpha = 0.6,
binwidth = 5) +
ggplot2::geom_density(alpha = 0.5) +
ggplot2::geom_vline(data = means, ggplot2::aes(xintercept = SickDays, color = Depression),
linetype = 'dashed') +
ggplot2::scale_fill_brewer(palette = 'Accent') +
ggplot2::scale_color_brewer(palette = 'Accent') +
ggplot2::theme_minimal() +
ggplot2::theme(text=ggplot2::element_text(family="Open Sans")) +
ggplot2::labs(title = "Density Sick Days Distribution by Diagnosis of Depression",
x = "Sick Days", y = 'Density')
return(hist)
}
depressionHist1 <- plotDepressionHist(depressionData, depressionMeans)
depressionHist2 <- plotDepressionHist(depressionDataPhd, depressionMeansPhd)
#---------------------------------------------------------------------------#
# Create Chronic Illness Plots #
#---------------------------------------------------------------------------#
# Create Marginal Frequency Bar Plot of Sick Days Indicator by Chronic Illness
chronicFreqBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Chronic, y = Freq, fill = SickDaysInd),
data = chronicFreqDf, stat = 'identity') +
ggplot2::geom_text(data = chronicFreqDf, ggplot2::aes(x = Chronic, y = pos,
label = Freq),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens', name = "Sick Days") +
ggplot2::labs(title = "Frequency Distribution",
x = "Diagnosis of Chronic Illness",
y = 'Number Sick Days')
#---------------------------------------------------------------------------#
# Create Marginal Proportion Bar Plot of Sick Days and Chronic Illness Indicators
chronicPropBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Chronic, y = pct, fill = SickDaysInd),
data = chronicPropDf, stat = 'identity') +
ggplot2::geom_text(data = chronicPropDf, ggplot2::aes(x = Chronic, y = pos,
label = paste0(pct,"%")),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens', name = "Sick Days") +
ggplot2::labs(title = "Proportional Distribution",
x = "Diagnosis of Chronic Illness",
y = 'Percentage Reporting Sick Days')
#---------------------------------------------------------------------------#
# Create violin plot of sick days grouped by diagnosis of chronic illness
chronicViolin <- ggplot2::ggplot(chronicData,
ggplot2::aes(x = Chronic, y = SickDays, fill = Chronic)) +
ggplot2::geom_violin(draw_quantiles = c(0.25, 0.5, 0.75)) +
ggplot2::labs(title = 'Violin Plot', fill = 'Chronic Illness') +
ggplot2::ylab('Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Create box plot of sick days grouped by diagnosis of chronic illness
chronicBox <- ggplot2::ggplot(chronicData,
ggplot2::aes(x = Chronic, y = SickDays, fill = Chronic)) +
ggplot2::geom_boxplot() +
ggplot2::labs(title = 'Box Plot', fill = 'Chronic Illness') +
ggplot2::ylab('Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Create histogram of restricted activity days by diagnosis of chronic illness
plotChronicHist <- function(plotData, means) {
# Create histogram of sick days grouped a diagnosis of chronic illness
hist <- ggplot2::ggplot(plotData,
ggplot2::aes(x=SickDays, color = Chronic, fill = Chronic)) +
ggplot2::geom_histogram(ggplot2::aes(y=..density..), position = "identity", alpha = 0.6,
binwidth = 5) +
ggplot2::geom_density(alpha = 0.5) +
ggplot2::geom_vline(data = means, ggplot2::aes(xintercept = SickDays, color = Chronic),
linetype = 'dashed') +
ggplot2::scale_fill_brewer(palette = 'Accent') +
ggplot2::scale_color_brewer(palette = 'Accent') +
ggplot2::theme_minimal() +
ggplot2::theme(text=ggplot2::element_text(family="Open Sans")) +
ggplot2::labs(title = "Sick Days by Chronic Diagnosis",
x = "Sick Days", y = 'Density')
return(hist)
}
chronicHist1 <- plotChronicHist(chronicData, chronicMeans)
chronicHist2 <- plotChronicHist(chronicDataPhd, chronicMeansPhd)
#---------------------------------------------------------------------------#
# Create Interaction Plots #
#---------------------------------------------------------------------------#
# Create Marginal Frequency Bar Plot of Sick Days Indicator by Depression
interactionFreqBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Depression, y = Freq, fill = Chronic),
data = interactionFreqDf, stat = 'identity') +
ggplot2::geom_text(data = interactionFreqDf, ggplot2::aes(x = Depression, y = pos,
label = Freq),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens',
name = "Diagnosis") +
ggplot2::labs(title = 'Frequency Distribution',
x = "Diagnosis of Depression",
y = 'Diagnosis of Chronic Illness')
#---------------------------------------------------------------------------#
# Create Marginal Proportion Bar Plot of Sick Days Indicator by Depression
interactionPropBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Depression, y = pct, fill = Chronic),
data = interactionPropDf, stat = 'identity') +
ggplot2::geom_text(data = interactionPropDf, ggplot2::aes(x = Depression, y = pos,
label = paste0(pct,"%")),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens',
name = "Diagnosis") +
ggplot2::labs(title = 'Proportional Distribution',
x = "Diagnosis of Depression",
y = 'Diagnosis of Chronic Illness')
#---------------------------------------------------------------------------#
# Plot histogram of restricted activity days by chronic illness and depression
plotinteractionHist <- function(plotData, means) {
# Create histogram of sick days by depression and chronic illness
hist <- ggplot2::ggplot(plotData,
ggplot2::aes(x=SickDays, fill = DepressionChronic)) +
ggplot2::geom_histogram(ggplot2::aes(y=..density..), position = "identity", alpha = 0.6,
binwidth = 5) +
ggplot2::geom_density(alpha = 0.5) +
ggplot2::geom_vline(data = means, ggplot2::aes(xintercept = SickDays),
linetype = 'dashed') +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::scale_color_brewer(palette = 'Greens') +
ggplot2::theme_minimal() +
ggplot2::theme(text=ggplot2::element_text(family="Open Sans")) +
ggplot2::labs(title = 'Sick Days by Diagnoses of Depression and Chronic Illness',
x = "Sick Days", y = 'Density')
return(hist)
}
interactionHist1 <- plotinteractionHist(interactionData, interactionMeans)
interactionHist2 <- plotinteractionHist(interactionDataPhd, interactionMeansPhd)
#---------------------------------------------------------------------------#
# Create Interaction Violin Plot
interactionViolin <- ggplot2::ggplot(interactionData,
ggplot2::aes(x = DepressionChronic, y = SickDays, fill = DepressionChronic)) +
ggplot2::geom_violin() +
ggplot2::labs(title = "Violin Plot",
x = 'Depression / Chronic Illness',
y = 'Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::stat_summary(fun.y = mean, geom = 'point') +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Create Interaction Box Plot
interactionBox <- ggplot2::ggplot(interactionData,
ggplot2::aes(x = DepressionChronic, y = SickDays, fill = DepressionChronic)) +
ggplot2::geom_boxplot() +
ggplot2::labs(title = "Box Plot",
x = 'Depression / Chronic Illness',
y = 'Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::stat_summary(fun.y = mean, geom = 'point') +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Create Mean Sick Days by Chronic Illness Bar Plot
allChronicBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Condition, y = Mean, fill = Condition),
data = allChronicStats, stat = 'identity') +
ggplot2::geom_text(data = allChronicStats, ggplot2::aes(x = Condition, y = Mean/2,
label = Mean),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
axis.ticks.x = ggplot2::element_blank(),
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'PRGn',
name = "Diagnosis") +
ggplot2::labs(title = 'Mean Sick Days by Condition',
y = 'Sick Days')
#---------------------------------------------------------------------------#
# Create model for interaction plot
interactionModel <- lm(SickDays ~ Depression + Chronic + Depression*Chronic,
data = interactionData)
#---------------------------------------------------------------------------#
# Conduct Tests #
#---------------------------------------------------------------------------#
# Conduct Wilcox and effect tests for depression on sick days
depressionTest <- wilcox.test(SickDays ~ Depression, data = depressionData,
conf.int = TRUE)
yesGroup <- depressionData %>% filter(Depression == 'Yes')
noGroup <- depressionData %>% filter(Depression == 'No')
depressionEffect <- ks.test(yesGroup$SickDays, noGroup$SickDays)
# Conduct Wilcox and effect tests for chronic illness on sick days
chronicTest <- wilcox.test(SickDays ~ Chronic, data = chronicData,
conf.int = TRUE)
yesGroup <- chronicData %>% filter(Chronic == 'Yes')
noGroup <- chronicData %>% filter(Chronic == 'No')
chronicEffect <- ks.test(yesGroup$SickDays, noGroup$SickDays)
# Conduct Kruskal and effect tests for depression n& chronic illness on sick days
interactionTest <- kruskal.test(SickDays ~ DepressionChronic, data = interactionData)
# Partition data by explanatory variable
group1 <- interactionData %>% filter(Depression == 'Yes' & Chronic == 'Yes') %>% select(SickDays)
group2 <- interactionData %>% filter(Depression == 'No' & Chronic == 'Yes') %>% select(SickDays)
group3 <- interactionData %>% filter(Depression == 'Yes' & Chronic == 'No') %>% select(SickDays)
group4 <- interactionData %>% filter(Depression == 'No' & Chronic == 'No') %>% select(SickDays)
# Conduct pairwise wilcox teste of equal medians
test1 <- wilcox.test(group1$SickDays, group2$SickDays, conf.int = TRUE)
test2 <- wilcox.test(group2$SickDays, group3$SickDays, conf.int = TRUE)
test3 <- wilcox.test(group3$SickDays, group4$SickDays, conf.int = TRUE)
# Calculate effect sizes for each
effect1 <- ks.test(group1$SickDays, group2$SickDays)
effect2 <- ks.test(group2$SickDays, group3$SickDays)
effect3 <- ks.test(group3$SickDays, group4$SickDays)
# Format Test Results
pairwise <- data.frame(Test = c(1:3),
A.Depression = c("Yes",'No', 'Yes'),
A.Chronic = c("Yes",'Yes', 'No'),
B.Depression = c("No", 'Yes', 'No'),
B.Chronic = c("Yes",'No', 'No'),
X2 = c(test1$statistic, test2$statistic, test3$statistic),
p = c(test1$p.value, test2$p.value, test3$p.value),
Estimate = c(test1$estimate, test2$estimate, test3$estimate),
Lower_CI = c(test1$conf.int[1], test2$conf.int[1], test3$conf.int[1]),
Upper_CI = c(test1$conf.int[2], test2$conf.int[2], test3$conf.int[2]),
Effect = c(effect1$statistic, effect2$statistic, effect3$statistic))
#---------------------------------------------------------------------------#
# Format and Return Results #
#---------------------------------------------------------------------------#
analysis <- list(
tables = list(
depressionFreqTbl = depressionFreqTbl,
depressionPropTbl = depressionPropTbl,
chronicTbl = chronicTbl,
chronicFreqTbl = chronicFreqTbl,
chronicPropTbl = chronicPropTbl,
interactionTbl = interactionTbl,
interactionFreqTbl = interactionFreqTbl,
interactionPropTbl = interactionPropTbl
),
dataFrames = list(
depressionData = depressionData,
chronicData = chronicData,
interactionData = interactionData
),
stats = list(
depression = depressionStats,
chronic = chronicStats,
interaction = interactionStats,
allChronic = allChronicStats
),
plots = list(
depressionFreqBar = depressionFreqBar,
depressionPropBar = depressionPropBar,
depressionHist1 = depressionHist1,
depressionHist2 = depressionHist2,
depressionViolin = depressionViolin,
depressionBox = depressionBox,
chronicFreqBar = chronicFreqBar,
chronicPropBar = chronicPropBar,
chronicHist1 = chronicHist1,
chronicHist2 = chronicHist2,
chronicViolin = chronicViolin,
chronicBox = chronicBox,
allChronic = allChronicBar,
interactionFreqBar = interactionFreqBar,
interactionPropBar = interactionPropBar,
interactionHist1 = interactionHist1,
interactionHist2 = interactionHist2,
interactionViolin = interactionViolin,
interactionBox = interactionBox
),
tests = list(
depressionTest = depressionTest,
depressionEffect = depressionEffect,
chronicTest = chronicTest,
chronicEffect = chronicEffect,
interactionTest = interactionTest,
interactionModel = interactionModel,
pairwise = pairwise
)
)
return(analysis)
}
## ---- end
DataScienceSalon/chronic documentation built on May 8, 2019, 11:47 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
#==============================================================================#
# analyzeRq3.R #
#==============================================================================#
#'
#' analyzeRq3
#'
#' \code{analyzeRq3(brfss)}{Performs analyses for research question #3,
#' To what degree does depression with co-occurring chronic illness affect
#' productivity, vis-a-vis that of diagnoses of depression or chronic illness
#' separately?}
#'
#' This function analyzes the effects of depression and chronic illness on the
#' response variable, the number of sick days reported in the 30 days
#' preceding the survey. This analysis is conducted in the following five
#' sections.
#'
#' @section Acquire Data:
#' This section extracts the number of sick days and the sick days range
#' variable, as well as the depression and chronic illnesses indicator
#' variables.
#'
#' @section Create Marginal and Joint Contingency Tables:
#' This section renders the tables that illuminate the marginal and joint
#' effects of depression and chronic illness on the number of sick days
#' reported. Descriptive statistics of sick days are also reported
#' for both depression and chronic illness, as well as their joint
#' effects on the number of sick days reported.
#'
#' @section Plots:
#' In this section, frequency and proportion bar plots, box plots, violin
#' plots and histograms are rendered to illuminate the distribution of
#' sick days vis-a-vis chronic illness and depression diagnoses.
#'
#' @section Statistical Tests:
#' In this section, the non-parametric Wilcoxin signed-rank and the
#' the Kolmogorov–Smirnov tests were administered to ascertain the
#' effect of depression and chronic illness on the number of sick days
#' reported. The Wilcox test compared the mean number of sick days
#' reported for both depression and chronic disease diagnoses for
#' statistically significant differences. The Komogorov-Smirnov tests
#' were conducted pairwise to reveal relative effects of depression
#' and chronic illness on sick days reported.
#'
#' @param brfss Data frame of preprocessed 2013 Behavioral Risk Factor
#' Surveillance System (BRFSS) telephone survey.
#'
#' @return analysis List containing frequency and proportion data, plots,
#' and statistical tests. The variables are as follows:
#' \itemize{
#' \itemize{Data Tables}{
#' \item{depressionFreqTbl}{Table of frequencies of depression diagnoses}
#' \item{depressionPropTbl}{Table of proportions of depression diagnoses}
#' \item{chronicFreqTbl}{Table of frequencies of chronic disease diagnoses}
#' \item{chronicPropTbl}{Table of proportions of chronic disease diagnoses}
#' \item{interactionFreqTbl}{Table of frequencies of the interaction between chronic illness and depression diagnoses}
#' \item{interactionPropTbl}{Table of proportions of the interaction between chronic illness and depression diagnoses}
#' }
#' \itemize{Data Frames}{
#' \item{depressionData}{Depression frequency data in data frame format}
#' \item{chronicData}{Chronic disease frequency data in data frame format}
#' \item{interactionData}{Interaction between depression and chronic illness frequency data in data frame format}
#' }
#' \itemize{Statistics}{
#' \item{depression}{Descriptive statistics for sick days vis-a-vis depression diagnoses}
#' \item{chronic}{Descriptive statistics for sick days vis-a-vis chronic illness diagnoses}
#' \item{interaction}{Descriptive statistics for sick days vis-a-vis the depression / chronic illness interaction variable}
#' \item{allChronic}{Descriptive statistics for sick days vis-a-vis a diagnoses of any chronic illness}
#' }
#' \itemize{Plots}{
#' \item{depressionFreqBar}{The depression frequency bar plot}
#' \item{depressionPropBar}{The depression proportion bar plot}
#' \item{depressionHist1}{A histogram of the distribution of sick days vis-a-vis a diagnosis of depression}
#' \item{depressionHist2}{A histogram of the distribution of one or more sick days reported vis-a-vis a diagnosis of depression}
#' \item{depressionViolin}{A violin plot of the distribution of sick days vis-a-vis a diagnosis of depression}
#' \item{depressionBox}{A box plot of the distribution of sick days vis-a-vis a diagnosis of depression}
#' \item{chronicFreqBar}{The chronic illness diagnosis frequency bar plot}
#' \item{chronicPropBar}{The chronic illness diagnosis proportion bar plot}
#' \item{chronicHist1}{A histogram of the distribution of sick days vis-a-vis a diagnosis of chronic illness}
#' \item{chronicHist2}{A histogram of the distribution of one or more sick days reported vis-a-vis a diagnosis of chronic illness}
#' \item{chronicViolin}{A violin plot of the distribution of sick days vis-a-vis a diagnosis of chronic illness}
#' \item{chronicBox}{A box plot of the distribution of sick days vis-a-vis a diagnosis of chronic illness}
#' \item{allChronic}{A box plot of the distribution of sick days vis-a-vis a diagnosis of of any chronic illness}
#' \item{interactionFreqBar}{The depression / chronic illness interaction variable frequency bar plot of sick }
#' \item{interactionPropBar}{The depression / chronic illness interaction variable proportion bar plot}
#' \item{interactionHist1}{A histogram of the distribution of sick days vis-a-vis the depression / chronic illness interaction variable}
#' \item{interactionHist2}{A histogram of the distribution of one or more sick days reported vis-a-vis the depression / chronic illness interaction variable}
#' \item{interactionViolin}{A violin plot of the distribution of sick days vis-a-vis the depression / chronic illness interaction variable}
#' \item{interactionBox}{A box plot of the distribution of sick days vis-a-vis the depression / chronic illness interaction variable}
#' \item{allinteraction}{A box plot of the distribution of sick days vis-a-vis the depression / chronic illness interaction variable}
#' }
#' \itemize{tests}{
#' \item{depressionTest}{The Wilcoxin ranked-sums test of equal means with and without depression}
#' \item{depressionEffect}{The Kolmogorov-Smirnov test of the effect of depression on sick days reported}
#' \item{chronicTest}{The Wilcoxin ranked-sums test of equal means with and without a diagnosis of chronic illness}
#' \item{chronicEffect}{The Kolmogorov-Smirnov test of the effect of chronic illness on sick days reported}
#' \item{interactionTest}{The Wilcoxin ranked-sums test of equal means vis-a-vis the depression / chronic illness interaction variable}
#' \item{interactionModel}{The linear model that relates sick days to the values of the interaction variable}
#' \item{pairwise}{The pairwise Kolmogorov-Smirnov test of the effect of depression and chronic illness on sick days reported}
#' }
#' }
#'
#'
#' @author John James, \email{jjames@@datasciencesalon.org}
#' @export
analyzeRq3 <- function(brfss) {
#---------------------------------------------------------------------------#
# Get Data #
#---------------------------------------------------------------------------#
# Get complete cases for interaction analysis and create interaction variable
interactionData <- brfss %>%
filter(!is.na(SickDays) & !is.na(Depression) & !is.na(Chronic) & SickDays < 32) %>%
select(SickDays, Depression, Chronic, SickDaysInd)
interactionData$DepressionChronic <-
interaction(interactionData$Depression, interactionData$Chronic)
interactionDataPhd <- interactionData %>% filter(SickDays > 0)
# Get depression data
depressionData <- brfss %>% filter(!is.na(SickDays) &
!is.na(Depression) & SickDays < 32) %>%
select(SickDays, Depression, SickDaysInd)
depressionDataPhd <- depressionData %>% filter(SickDays > 0)
# Get chronic data
chronicData <- brfss %>% filter(!is.na(SickDays) &
!is.na(Chronic) & SickDays < 32) %>%
select(SickDays, Chronic, SickDaysInd)
chronicDataPhd <- chronicData %>% filter(SickDays > 0)
# Get data for all chronic disease, including depression
allChronicData <- brfss %>% filter(!is.na(SickDays) & Depression == 'No' &
SickDays < 32 &
!is.na('Heart Attack') & !is.na(Chronic) &
Chronic == 'Yes' &
!is.na('Angina Or Coronary Heart Disease') &
!is.na(Stroke) & !is.na(Asthma) &
!is.na('Skin Cancer') & !is.na('Cancer (Other)') &
!is.na(COPD) & !is.na(Arthritis) &
!is.na('Kidney Disease') & !is.na(Diabetes)) %>%
select(SickDays, Depression, SickDaysInd, 'Heart Attack',
'Angina Or Coronary Heart Disease', Stroke, Asthma, 'Skin Cancer',
'Cancer (Other)', COPD, Arthritis, Diabetes, 'Kidney Disease')
#---------------------------------------------------------------------------#
# Create Depression Marginal Contingency Tables #
#---------------------------------------------------------------------------#
# Create Tables for Analysis of Effect of Depression on Productivity #
#---------------------------------------------------------------------------#
# Create 2-Way Marginal Contingency Tables on Depression and Sick Days
depressionTbl <- with(depressionData, table(Depression, SickDaysInd))
depressionFreqTbl <- ftable(depressionTbl)
depressionFreqDf <- as.data.frame(depressionFreqTbl)
depressionPropTbl <- ftable(prop.table(depressionTbl, 1))
depressionPropDf <- as.data.frame(depressionPropTbl)
# Format 2-Way Depression Marginal Frequency Contingency Tables for plotting
depressionFreqDf <- depressionFreqDf %>% arrange(Depression,
desc(SickDaysInd)) %>%
group_by(Depression) %>%
mutate(cumFreq = cumsum(Freq), pos = cumFreq - 0.5 * Freq)
# Format 2-Way Depression Marginal Proportion Contingency Tables for plotting
depressionPropDf <- depressionPropDf %>% arrange(Depression,
desc(SickDaysInd)) %>%
group_by(Depression) %>%
mutate(pct = round(Freq * 100, 0), cumPct = cumsum(pct),
pos = cumPct - 0.5 * pct)
#---------------------------------------------------------------------------#
# Create Chronic Illness Marginal Contingency Tables #
#---------------------------------------------------------------------------#
# Create Tables for Analysis of Effect of Chronic Illness on Productivity #
#---------------------------------------------------------------------------#
# Create 2-Way Marginal Contingency Tables on Chronic Ilness and Sick Days
chronicTbl <- with(chronicData, table(Chronic, SickDaysInd))
chronicFreqTbl <- ftable(chronicTbl)
chronicFreqDf <- as.data.frame(chronicFreqTbl)
chronicPropTbl <- ftable(prop.table(chronicTbl, 1))
chronicPropDf <- as.data.frame(chronicPropTbl)
# Format 2-Way Chronic Marginal Frequency Contingency Tables for plotting
chronicFreqDf <- chronicFreqDf %>% arrange(Chronic,
desc(SickDaysInd)) %>%
group_by(Chronic) %>%
mutate(cumFreq = cumsum(Freq), pos = cumFreq - 0.5 * Freq)
# Format 2-Way Chronic Marginal Proportional Contingency Tables for plotting
chronicPropDf <- chronicPropDf %>% arrange(Chronic,
desc(SickDaysInd)) %>%
group_by(Chronic) %>%
mutate(pct = round(Freq * 100, 0), cumPct = cumsum(pct),
pos = cumPct - 0.5 * pct)
#---------------------------------------------------------------------------#
# Create Depression and Chronic Illness Contingency Tables #
#---------------------------------------------------------------------------#
# Create the 2X2 Depression and Chronic Illnesses Interaction Tables #
#---------------------------------------------------------------------------#
# Create Joint Contingency Tables on Depression and Chronic Illness
interactionTbl <- with(interactionData, table(Depression, Chronic, SickDaysInd))
interactionFreqTbl <- ftable(interactionTbl)
interactionFreqDf <- as.data.frame(interactionFreqTbl)
interactionPropTbl <- ftable(prop.table(interactionTbl, 1))
interactionPropDf <- as.data.frame(interactionPropTbl)
# Format 2-Way Depression Marginal Frequency Contingency Tables for plotting
interactionFreqDf <- interactionFreqDf %>% arrange(Depression, Chronic,
desc(SickDaysInd)) %>%
group_by(Depression) %>%
mutate(cumFreq = cumsum(Freq), pos = cumFreq - 0.5 * Freq)
# Format 2-Way Depression Marginal Proportion Contingency Tables for plotting
interactionPropDf <- interactionPropDf %>% arrange(Depression, Chronic,
desc(SickDaysInd)) %>%
group_by(Depression) %>%
mutate(pct = round(Freq * 100, 0), cumPct = cumsum(pct),
pos = cumPct - 0.5 * pct)
#---------------------------------------------------------------------------#
# Get Summary Statistics #
#---------------------------------------------------------------------------#
# Get mean of restricted activity days by depression
depressionMeans <- aggregate(SickDays ~ Depression, interactionData, mean)
depressionMeansPhd <- aggregate(SickDays ~ Depression, interactionDataPhd, mean)
# Get mean of restricted activity days by chronic
chronicMeans <- aggregate(SickDays ~ Chronic, interactionData, mean)
chronicMeansPhd <- aggregate(SickDays ~ Chronic, interactionDataPhd, mean)
# Get mean of restricted activity days by depression and chronic illness
interactionMeans <- aggregate(SickDays ~ Depression + Chronic, interactionData, mean)
interactionMeansPhd <- aggregate(SickDays ~ Depression + Chronic, interactionDataPhd, mean)
#---------------------------------------------------------------------------#
# Summary stats function
getStats <- function(theData, varName, eVar, value) {
stats <- psych::describe(theData)
stats <- data.frame(stats, row.names = NULL)
ci <- pastecs::stat.desc(theData)[11,1]
ds <- data.frame(Variable = varName,
Value = value,
N = stats$n,
Min = stats$min,
Lower = quantile(theData, .25, na.rm = TRUE),
Median = stats$median,
Mode = getMode(as.vector(theData$SickDays)),
Mean = round(stats$mean, 2),
CI = round(ci, 3),
Upper = quantile(theData, .75, na.rm = TRUE),
Max = stats$max,
Range = stats$range,
Total = sum(theData, na.rm = TRUE),
SD = round(stats$sd, 2),
SE = round(stats$se, 3),
Skew = round(stats$skew, 2),
Kurtosis = round(stats$kurtosis, 2))
names(ds)[names(ds) == 'Value'] <- eVar
return(ds)
}
#---------------------------------------------------------------------------#
# Get summary stats of sick days by depression, chronic illness, and interaction
values <- list('Yes', 'No')
depressionStats <-
data.table::rbindlist(lapply(seq_along(values), function(x) {
getStats(depressionData %>% filter(Depression == values[[x]])
%>% select(SickDays), 'SickDays', 'Depression', values[[x]])
}))
chronicStats <-
data.table::rbindlist(lapply(seq_along(values), function(x) {
getStats(chronicData %>% filter(Chronic == values[[x]])
%>% select(SickDays), 'SickDays', 'Chronic', values[[x]])
}))
# Chronic Illness Stats
depStats <- getStats(depressionData %>% filter(depressionData[,2] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Depression')
haStats <- getStats(allChronicData %>% filter(allChronicData[,4] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Heart Attack')
anginaStats <- getStats(allChronicData %>% filter(allChronicData[,5] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Angina or CHD')
strokeStats <- getStats(allChronicData %>% filter(allChronicData[,6] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Stroke')
asthmaStats <- getStats(allChronicData %>% filter(allChronicData[,7] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Asthma')
skinStats <- getStats(allChronicData %>% filter(allChronicData[,8] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Skin Cancer')
cancerStats <- getStats(allChronicData %>% filter(allChronicData[,9] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Cancer (Other)')
copdStats <- getStats(allChronicData %>% filter(allChronicData[,10] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'COPD')
arthritisStats <- getStats(allChronicData %>% filter(allChronicData[,11] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Arthritis')
diabetesStats <- getStats(allChronicData %>% filter(allChronicData[,12] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Diabetes')
kidneyStats <- getStats(allChronicData %>% filter(allChronicData[,13] == 'Yes')
%>% select(SickDays), 'SickDays', 'Condition', 'Kidney Disease')
allChronicStats <- rbind(depStats, haStats, anginaStats, strokeStats, asthmaStats, skinStats,
cancerStats, copdStats, arthritisStats, diabetesStats,
kidneyStats)
allChronicStats <- allChronicStats[order(-allChronicStats$Mean),]
values <- list('Yes.Yes', 'No.Yes', 'Yes.No', 'No.No')
interactionStats <-
data.table::rbindlist(lapply(seq_along(values), function(x) {
getStats(interactionData %>% filter(DepressionChronic == values[[x]])
%>% select(SickDays), 'SickDays', 'DepressionChronic', values[[x]])
}))
#---------------------------------------------------------------------------#
# Create Depression Plots #
#---------------------------------------------------------------------------#
# Create Marginal Frequency Bar Plot of Sick Days Indicator by Depression
depressionFreqBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Depression, y = Freq, fill = SickDaysInd),
data = depressionFreqDf, stat = 'identity') +
ggplot2::geom_text(data = depressionFreqDf, ggplot2::aes(x = Depression, y = pos,
label = Freq),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens',
name = "Sick Days") +
ggplot2::labs(title = "Frequency Distribution",
x = "Diagnosis of Depression",
y = 'Number Reporting Sick Days')
#---------------------------------------------------------------------------#
# Create Marginal Proportion Bar Plot of Sick Days Indicator by Depression
depressionPropBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Depression, y = pct, fill = SickDaysInd),
data = depressionPropDf, stat = 'identity') +
ggplot2::geom_text(data = depressionPropDf, ggplot2::aes(x = Depression, y = pos,
label = paste0(pct,"%")),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens', name = "Sick Days") +
ggplot2::labs(title = "Proportional Distribution",
x = "Diagnosis of Depression",
y = 'Percentage Reporting Sick Days')
#---------------------------------------------------------------------------#
# Create violin of sick days grouped by diagnosis of depression
depressionViolin <- ggplot2::ggplot(depressionData,
ggplot2::aes(x = Depression, y = SickDays, fill = Depression)) +
ggplot2::geom_violin(draw_quantiles = c(0.25, 0.5, 0.75)) +
ggplot2::labs(title = 'Violin Plot',fill = 'Depression') +
ggplot2::ylab('Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Create box of sick days grouped by diagnosis of depression
depressionBox <- ggplot2::ggplot(depressionData,
ggplot2::aes(x = Depression, y = SickDays, fill = Depression)) +
ggplot2::geom_boxplot() +
ggplot2::labs(title = 'Box Plot', fill = 'Depression') +
ggplot2::ylab('Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Plot a histogram of restricted activity days by depression diagnosis
plotDepressionHist <- function(plotData, means) {
# Create histogram of sick days grouped a diagnosis of depression
hist <- ggplot2::ggplot(plotData,
ggplot2::aes(x=SickDays, color = Depression, fill = Depression)) +
ggplot2::geom_histogram(ggplot2::aes(y=..density..), position = "identity", alpha = 0.6,
binwidth = 5) +
ggplot2::geom_density(alpha = 0.5) +
ggplot2::geom_vline(data = means, ggplot2::aes(xintercept = SickDays, color = Depression),
linetype = 'dashed') +
ggplot2::scale_fill_brewer(palette = 'Accent') +
ggplot2::scale_color_brewer(palette = 'Accent') +
ggplot2::theme_minimal() +
ggplot2::theme(text=ggplot2::element_text(family="Open Sans")) +
ggplot2::labs(title = "Density Sick Days Distribution by Diagnosis of Depression",
x = "Sick Days", y = 'Density')
return(hist)
}
depressionHist1 <- plotDepressionHist(depressionData, depressionMeans)
depressionHist2 <- plotDepressionHist(depressionDataPhd, depressionMeansPhd)
#---------------------------------------------------------------------------#
# Create Chronic Illness Plots #
#---------------------------------------------------------------------------#
# Create Marginal Frequency Bar Plot of Sick Days Indicator by Chronic Illness
chronicFreqBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Chronic, y = Freq, fill = SickDaysInd),
data = chronicFreqDf, stat = 'identity') +
ggplot2::geom_text(data = chronicFreqDf, ggplot2::aes(x = Chronic, y = pos,
label = Freq),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens', name = "Sick Days") +
ggplot2::labs(title = "Frequency Distribution",
x = "Diagnosis of Chronic Illness",
y = 'Number Sick Days')
#---------------------------------------------------------------------------#
# Create Marginal Proportion Bar Plot of Sick Days and Chronic Illness Indicators
chronicPropBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Chronic, y = pct, fill = SickDaysInd),
data = chronicPropDf, stat = 'identity') +
ggplot2::geom_text(data = chronicPropDf, ggplot2::aes(x = Chronic, y = pos,
label = paste0(pct,"%")),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens', name = "Sick Days") +
ggplot2::labs(title = "Proportional Distribution",
x = "Diagnosis of Chronic Illness",
y = 'Percentage Reporting Sick Days')
#---------------------------------------------------------------------------#
# Create violin plot of sick days grouped by diagnosis of chronic illness
chronicViolin <- ggplot2::ggplot(chronicData,
ggplot2::aes(x = Chronic, y = SickDays, fill = Chronic)) +
ggplot2::geom_violin(draw_quantiles = c(0.25, 0.5, 0.75)) +
ggplot2::labs(title = 'Violin Plot', fill = 'Chronic Illness') +
ggplot2::ylab('Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Create box plot of sick days grouped by diagnosis of chronic illness
chronicBox <- ggplot2::ggplot(chronicData,
ggplot2::aes(x = Chronic, y = SickDays, fill = Chronic)) +
ggplot2::geom_boxplot() +
ggplot2::labs(title = 'Box Plot', fill = 'Chronic Illness') +
ggplot2::ylab('Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Create histogram of restricted activity days by diagnosis of chronic illness
plotChronicHist <- function(plotData, means) {
# Create histogram of sick days grouped a diagnosis of chronic illness
hist <- ggplot2::ggplot(plotData,
ggplot2::aes(x=SickDays, color = Chronic, fill = Chronic)) +
ggplot2::geom_histogram(ggplot2::aes(y=..density..), position = "identity", alpha = 0.6,
binwidth = 5) +
ggplot2::geom_density(alpha = 0.5) +
ggplot2::geom_vline(data = means, ggplot2::aes(xintercept = SickDays, color = Chronic),
linetype = 'dashed') +
ggplot2::scale_fill_brewer(palette = 'Accent') +
ggplot2::scale_color_brewer(palette = 'Accent') +
ggplot2::theme_minimal() +
ggplot2::theme(text=ggplot2::element_text(family="Open Sans")) +
ggplot2::labs(title = "Sick Days by Chronic Diagnosis",
x = "Sick Days", y = 'Density')
return(hist)
}
chronicHist1 <- plotChronicHist(chronicData, chronicMeans)
chronicHist2 <- plotChronicHist(chronicDataPhd, chronicMeansPhd)
#---------------------------------------------------------------------------#
# Create Interaction Plots #
#---------------------------------------------------------------------------#
# Create Marginal Frequency Bar Plot of Sick Days Indicator by Depression
interactionFreqBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Depression, y = Freq, fill = Chronic),
data = interactionFreqDf, stat = 'identity') +
ggplot2::geom_text(data = interactionFreqDf, ggplot2::aes(x = Depression, y = pos,
label = Freq),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens',
name = "Diagnosis") +
ggplot2::labs(title = 'Frequency Distribution',
x = "Diagnosis of Depression",
y = 'Diagnosis of Chronic Illness')
#---------------------------------------------------------------------------#
# Create Marginal Proportion Bar Plot of Sick Days Indicator by Depression
interactionPropBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Depression, y = pct, fill = Chronic),
data = interactionPropDf, stat = 'identity') +
ggplot2::geom_text(data = interactionPropDf, ggplot2::aes(x = Depression, y = pos,
label = paste0(pct,"%")),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'Greens',
name = "Diagnosis") +
ggplot2::labs(title = 'Proportional Distribution',
x = "Diagnosis of Depression",
y = 'Diagnosis of Chronic Illness')
#---------------------------------------------------------------------------#
# Plot histogram of restricted activity days by chronic illness and depression
plotinteractionHist <- function(plotData, means) {
# Create histogram of sick days by depression and chronic illness
hist <- ggplot2::ggplot(plotData,
ggplot2::aes(x=SickDays, fill = DepressionChronic)) +
ggplot2::geom_histogram(ggplot2::aes(y=..density..), position = "identity", alpha = 0.6,
binwidth = 5) +
ggplot2::geom_density(alpha = 0.5) +
ggplot2::geom_vline(data = means, ggplot2::aes(xintercept = SickDays),
linetype = 'dashed') +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::scale_color_brewer(palette = 'Greens') +
ggplot2::theme_minimal() +
ggplot2::theme(text=ggplot2::element_text(family="Open Sans")) +
ggplot2::labs(title = 'Sick Days by Diagnoses of Depression and Chronic Illness',
x = "Sick Days", y = 'Density')
return(hist)
}
interactionHist1 <- plotinteractionHist(interactionData, interactionMeans)
interactionHist2 <- plotinteractionHist(interactionDataPhd, interactionMeansPhd)
#---------------------------------------------------------------------------#
# Create Interaction Violin Plot
interactionViolin <- ggplot2::ggplot(interactionData,
ggplot2::aes(x = DepressionChronic, y = SickDays, fill = DepressionChronic)) +
ggplot2::geom_violin() +
ggplot2::labs(title = "Violin Plot",
x = 'Depression / Chronic Illness',
y = 'Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::stat_summary(fun.y = mean, geom = 'point') +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Create Interaction Box Plot
interactionBox <- ggplot2::ggplot(interactionData,
ggplot2::aes(x = DepressionChronic, y = SickDays, fill = DepressionChronic)) +
ggplot2::geom_boxplot() +
ggplot2::labs(title = "Box Plot",
x = 'Depression / Chronic Illness',
y = 'Sick Days') +
ggplot2::theme_minimal() +
ggplot2::scale_fill_brewer(palette = 'Greens', direction = -1) +
ggplot2::stat_summary(fun.y = mean, geom = 'point') +
ggplot2::theme(text = ggplot2::element_text(family="Open Sans"),
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
legend.position = 'bottom')
#---------------------------------------------------------------------------#
# Create Mean Sick Days by Chronic Illness Bar Plot
allChronicBar <- ggplot2::ggplot() +
ggplot2::geom_bar(ggplot2::aes(x = Condition, y = Mean, fill = Condition),
data = allChronicStats, stat = 'identity') +
ggplot2::geom_text(data = allChronicStats, ggplot2::aes(x = Condition, y = Mean/2,
label = Mean),
colour="black", family="Tahoma", size=4) +
ggplot2::theme_minimal() +
ggplot2::theme(legend.position="bottom",
axis.title.x = ggplot2::element_blank(),
axis.text.x = ggplot2::element_blank(),
axis.ticks.x = ggplot2::element_blank(),
text=ggplot2::element_text(family="Open Sans")) +
ggplot2::scale_fill_brewer(palette = 'PRGn',
name = "Diagnosis") +
ggplot2::labs(title = 'Mean Sick Days by Condition',
y = 'Sick Days')
#---------------------------------------------------------------------------#
# Create model for interaction plot
interactionModel <- lm(SickDays ~ Depression + Chronic + Depression*Chronic,
data = interactionData)
#---------------------------------------------------------------------------#
# Conduct Tests #
#---------------------------------------------------------------------------#
# Conduct Wilcox and effect tests for depression on sick days
depressionTest <- wilcox.test(SickDays ~ Depression, data = depressionData,
conf.int = TRUE)
yesGroup <- depressionData %>% filter(Depression == 'Yes')
noGroup <- depressionData %>% filter(Depression == 'No')
depressionEffect <- ks.test(yesGroup$SickDays, noGroup$SickDays)
# Conduct Wilcox and effect tests for chronic illness on sick days
chronicTest <- wilcox.test(SickDays ~ Chronic, data = chronicData,
conf.int = TRUE)
yesGroup <- chronicData %>% filter(Chronic == 'Yes')
noGroup <- chronicData %>% filter(Chronic == 'No')
chronicEffect <- ks.test(yesGroup$SickDays, noGroup$SickDays)
# Conduct Kruskal and effect tests for depression n& chronic illness on sick days
interactionTest <- kruskal.test(SickDays ~ DepressionChronic, data = interactionData)
# Partition data by explanatory variable
group1 <- interactionData %>% filter(Depression == 'Yes' & Chronic == 'Yes') %>% select(SickDays)
group2 <- interactionData %>% filter(Depression == 'No' & Chronic == 'Yes') %>% select(SickDays)
group3 <- interactionData %>% filter(Depression == 'Yes' & Chronic == 'No') %>% select(SickDays)
group4 <- interactionData %>% filter(Depression == 'No' & Chronic == 'No') %>% select(SickDays)
# Conduct pairwise wilcox teste of equal medians
test1 <- wilcox.test(group1$SickDays, group2$SickDays, conf.int = TRUE)
test2 <- wilcox.test(group2$SickDays, group3$SickDays, conf.int = TRUE)
test3 <- wilcox.test(group3$SickDays, group4$SickDays, conf.int = TRUE)
# Calculate effect sizes for each
effect1 <- ks.test(group1$SickDays, group2$SickDays)
effect2 <- ks.test(group2$SickDays, group3$SickDays)
effect3 <- ks.test(group3$SickDays, group4$SickDays)
# Format Test Results
pairwise <- data.frame(Test = c(1:3),
A.Depression = c("Yes",'No', 'Yes'),
A.Chronic = c("Yes",'Yes', 'No'),
B.Depression = c("No", 'Yes', 'No'),
B.Chronic = c("Yes",'No', 'No'),
X2 = c(test1$statistic, test2$statistic, test3$statistic),
p = c(test1$p.value, test2$p.value, test3$p.value),
Estimate = c(test1$estimate, test2$estimate, test3$estimate),
Lower_CI = c(test1$conf.int[1], test2$conf.int[1], test3$conf.int[1]),
Upper_CI = c(test1$conf.int[2], test2$conf.int[2], test3$conf.int[2]),
Effect = c(effect1$statistic, effect2$statistic, effect3$statistic))
#---------------------------------------------------------------------------#
# Format and Return Results #
#---------------------------------------------------------------------------#
analysis <- list(
tables = list(
depressionFreqTbl = depressionFreqTbl,
depressionPropTbl = depressionPropTbl,
chronicTbl = chronicTbl,
chronicFreqTbl = chronicFreqTbl,
chronicPropTbl = chronicPropTbl,
interactionTbl = interactionTbl,
interactionFreqTbl = interactionFreqTbl,
interactionPropTbl = interactionPropTbl
),
dataFrames = list(
depressionData = depressionData,
chronicData = chronicData,
interactionData = interactionData
),
stats = list(
depression = depressionStats,
chronic = chronicStats,
interaction = interactionStats,
allChronic = allChronicStats
),
plots = list(
depressionFreqBar = depressionFreqBar,
depressionPropBar = depressionPropBar,
depressionHist1 = depressionHist1,
depressionHist2 = depressionHist2,
depressionViolin = depressionViolin,
depressionBox = depressionBox,
chronicFreqBar = chronicFreqBar,
chronicPropBar = chronicPropBar,
chronicHist1 = chronicHist1,
chronicHist2 = chronicHist2,
chronicViolin = chronicViolin,
chronicBox = chronicBox,
allChronic = allChronicBar,
interactionFreqBar = interactionFreqBar,
interactionPropBar = interactionPropBar,
interactionHist1 = interactionHist1,
interactionHist2 = interactionHist2,
interactionViolin = interactionViolin,
interactionBox = interactionBox
),
tests = list(
depressionTest = depressionTest,
depressionEffect = depressionEffect,
chronicTest = chronicTest,
chronicEffect = chronicEffect,
interactionTest = interactionTest,
interactionModel = interactionModel,
pairwise = pairwise
)
)
return(analysis)
}
## ---- end
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.