Project 7: Emotional and Personal Health
Uriel Serna Haro and Michael Stauffer
4/11/2021
Abstract
In this project, we will explore the relationships among various network-based factors and the mental health of individuals . Different social network analysis approaches will be implemented to see if these factors can help us explain the trends and effects of mental illness among the networks.
Introduction
It has been established through research that the human species relies on companionship to lead a happy and healthy life. The biggest influences of our happiness is strongly associated with the ties formed among our family and friends. One way that our social network correlates with our subjective well-being is the respect and positive feedback from our network that boosts our self-esteem (Atchley, 1991). By identifying which ties have the most impact on the mental well-being of a network’s actors, we can suggest changes to the actors’ ties among family and friends to help improve their overall mental health. In this study, we well specifically look into the cycles of a network to determine if there are any redundant ties that have the greatest influence, negative or positive, among the actors. A cycle is a path consisting of at least three different nodes that start and end wit the same node. These cycles help identify how effective dissemination is among the network. The redundant ties within these cycles will help us determine whether those paths have a positive or negative effect on the actors as a whole (Healy, 2004). R libraries used for assignment:
Project Specification
The project seeks to analyze emotional health networks and the impact social factors have on people's mental well-being. A network will be examined with regard to a specific dataset and a set of node connections. The impact of mental health between connections of individuals of the network will then be examined visually. With centrality measures calculated, the most important nodes and their effect on others can be assessed and further determinations made. The project will then apply the network to statistical tests to identify correlations between mental health and social factors. Overall, conclusions can be drawn to further understand the impact of different characteristics on the overall mental well-being of the social network. Problem Analysis and Solution Design
The project will use an example emotional health dataset to identify the impact of mental and social factors on people's lives. The dataset will be used as a network by combining the health with a set of contrived node connections to build an example health network. The factors of the original dataset will be then be assessed from three perspectives visually. Each visual graph will compare the network with regard to two key components of persons of the original dataset. Centrality measures will then be drawn for the overall network to define the most important, most central nodes in the network. With visuals and metrics, further analysis and recommendations can be made with regard to the network. Then, the network will be applied to three statistical tests using QAP linear regression. These tests will seek to understand the correlation between four social factors with one mental health characteristic. Comparing similarity adjacency matrices of the factors, the assessment of how well each social factor describes the mental health factor can be assessed. In the end, conclusions will be sought, summarizing the overall impact of social characteristics on the mental health of persons of the network.# Import necessary libraries
library(statnet)
library(GGally)
# Ensure "igraph" does not conflict with "network" (Referenced kohske (2020))
library(igraph)
detach("package:igraph", unload=TRUE)Background Research
The following expression will be used to mitigate the influence of cyclic and acyclic node pairs on the characteristic path length of the graph: 
Where the total of all the shortest distances eithin all cycles found within the network are divided by H(G), the set of all cyclic nodes (Zhou, Liang, Zhao, et Zhang, 2018). The data used for this assignment is based on the answers from a survey conducted among individuals in Bangladesh. The survey asks various questions relating to the participants’ emotional issues and their drug use among other details of their lifestyles such as who they live and associate with (Protik, 2021). To make it more suitable for analysis, responses were transformed to binary values. A separate data set, adj_network2, holds an adjacency matrix of the interactions among a sample of the participants. Here is a sample of the data:
# Import dataset
dt = read.csv("drugAddiction_Revised.csv", header = TRUE, row.names = 1, check.names =FALSE)
# Add "ID" column to data frame (Referenced Simpson (2013))
dt <- cbind("ID" = 1:211, dt)
dt$ID <- c(1:211)
# Define column names
colnames(dt) <- c("ID", "Age", "Gender", "Live_with", "Conflict_with_law", "Spend_most_time", "Failure_in_life", "Mental_problem", "Suicidal_thoughts", "Family_relationship", "Number_of_friends", "Case_in_court", "Smoking", "Acquaintance")
# Display head of data
head(dt, 10)## ID Age Gender Live_with Conflict_with_law Spend_most_time Failure_in_life
## 1 1 2 0 0 0 0 1
## 2 2 2 0 1 1 1 1
## 3 3 1 0 1 1 1 1
## 4 4 2 0 1 0 1 1
## 5 5 2 0 1 0 0 0
## 6 6 2 0 1 0 0 0
## 7 7 2 0 1 0 0 0
## 8 8 2 0 1 0 1 1
## 9 9 2 0 1 0 1 1
## 10 10 2 0 0 0 0 1
## Mental_problem Suicidal_thoughts Family_relationship Number_of_friends
## 1 1 1 2 2
## 2 1 0 2 2
## 3 1 1 1 2
## 4 1 0 1 2
## 5 1 0 2 2
## 6 1 0 2 2
## 7 1 0 2 2
## 8 1 0 1 2
## 9 1 0 1 2
## 10 1 1 0 1
## Case_in_court Smoking Acquaintance
## 1 0 1 210
## 2 0 1 2
## 3 1 1 171
## 4 0 1 67
## 5 0 1 69
## 6 0 1 NA
## 7 0 1 83
## 8 0 1 20
## 9 0 1 60
## 10 0 1 15# Import network connections dataset
el = read.csv("adj_network2.csv", header = TRUE, row.names = 1, check.names =FALSE)
mat <- as.matrix(el)General visualization of age and gender among participants
The purpose for the first visualization below is to get a glimpse of the general demographics of the participants in the study. The key for this graph is:
gender - 0 = male, 1 = female
age - 1 = 15 to 22, 2 = 22 to 35From this graph we can see that the actors in the network are predominantly males aged 22 to 35. This network is composed of 67% males and 33% females.
# Initialize network nw
nw <- network(mat, directed = TRUE)
# Define ID attribute
att <- data.frame('ID' = as.character(dt$ID))
# Set gender attribute as vertex
att$Gender <- as.character(dt$Gender[match(att$ID, dt$ID)])
set.vertex.attribute(nw, 'Gender', as.character(att$Gender))
# Set age attribute as vertex
att$Age <- dt$Age[match(att$ID, dt$ID)]
set.vertex.attribute(nw, 'Age', att$Age)
# Define graph color palette
colours <- c('#f0f9e8','#ccebc5','#a8ddb5','#7bccc4')
names(colours) <- sort(unique(att$Age))
# Print network graph
ggnet2(nw,
label = TRUE,
label.size = 3,
node.shape = 'Gender',
node.color = 'Age',
palette = colours)
Visualization #2, Mental Health Issues and Living Conditions
The next graph of the network visualizes the effects of the participants’ living conditions correlate to their mental health. The legend for this graph is:
live With - 0 = alone, 1 = with family/relatives Mental Problem - 0 = no, 1 = yes
We can see that the majority of the participants, 62%, live at home with family while 38% live alone. The mental health issues among the individuals include depression, anxiety, tension, and anger. These various health issues do not seem to be directly correlated with ones living conditions, therefore analysis of other factors would be required.
# Initialize network nw2
nw2 <- network(mat, directed = TRUE)
# Define ID attribute
att2 <- data.frame('ID' = as.character(dt$ID))
# Set live with attribute as vertex
att2$LiveWith <- as.character(dt$Live_with[match(att2$ID, dt$ID)])
set.vertex.attribute(nw2, 'LiveWith', as.character(att2$LiveWith))
# Set mental problem attribute as vertex
att2$MentalProblem <- dt$Mental_problem[match(att2$ID, dt$ID)]
set.vertex.attribute(nw2, 'MentalProblem', as.character(att2$MentalProblem))
# Define graph color palette
colours2 <- c('#ecedd8','#d93d1e')
names(colours2) <- sort(unique(dt$Mental_problem))
# Print network graph
ggnet2(nw2,
label = TRUE,
label.size = 3,
node.shape = 'MentalProblem',
node.color = 'LiveWith',
palette = colours2)
Redundant Ties
From the graph, we can clearly see that participant number one lives by himself and suffers from one of the mental issues mentioned earlier and they are also connected to more participants with similar mental health conditions. To improve the overall mental health of the network, it would be wise to recommend changing this individuals living conditions to stay with friends or family to see if it would affect their overall mental health by establishing a form of companionship at home.
visualization #3 for Suicidal Thoughts and Who They Spend Most of Their Time With
Suicidal thoughts can be very dangerous among people with mental health issues as they have been strongly linked to depression in the past. To assist in our hypothesis that those individuals who are surrounded the most by family and friends tend to live a happier lifestyle, we compared the survey answers to whether participants encounter suicidal thoughts and who they spend their time with the most. The legend for this graph is:
Suicidal Thoughts - 0 = no, 1 = yes Who They spend Time with the Most - 0 = alone, 1 = friends, 2 = family
Looking at the visualization, we can easily see that those who spend most of their time with friends and family have fewer instances of suicidal thoughts than those who live alone, regardless of who they spend associate with within their network.
# Initialize network nw3
nw3 <- network(mat, directed = TRUE)
# Define ID attribute
att3 <- data.frame('ID' = as.character(att$ID))
# Set suicidal thoughts attribute as vertex
att3$Suicidal <- as.character(dt$Suicidal_thoughts[match(att3$ID, dt$ID)])
set.vertex.attribute(nw3, 'SuicidalThoughts', as.character(att3$Suicidal))
# Set spend time with attribute as vertex
att3$SpendTimeWith <- dt$Spend_most_time[match(att3$ID, dt$ID)]
set.vertex.attribute(nw3, 'SpendTimeWIth', as.character(att3$SpendTimeWith))
# Define graph color palette
colours3 <- c('#1a0624','#ded3e3','#a66fbf')
names(colours3) <- sort(unique(att3$SpendTimeWith))
# Print network graph
ggnet2(nw3,
label = TRUE,
label.size = 3,
label.color = "red",
node.shape = 'SuicidalThoughts',
node.color = 'SpendTimeWIth',
palette = colours3)
Redundant Ties
The redundant ties in this graph are also associated with those participants who encounter suicidal thoughts and spend the majority of their time by themselves (black triangles). These nodes also seem to be connected more to other individuals who also have suicidal thoughts, regardless of who they spend time with themselves. It would be strongly recommended to these individuals to associate themselves with the actors of the network that do not encounter suicidal thoughts and who do not spend their time alone to establish stronger lines of communication which may be beneficial in preventing those thoughts from escalating any further.
Centrality Measures and Cycles Evaluation
To identify the actors of our network that have the most influence among others, we applied degree, betweenness, and centrality measures to reveal which participants would be ideal to spread support to those whose mental well-being is in danger. Nodes 1, 33 and 57 ranked highest in all three measures which identifies them as the most influential actors of the network and can spread information most efficiently.
# Degree
degree(el)## [1] 28 1 1 2 0 1 2 1 2 0 0 0 0 0 0 0 0 0 5 0 0 0 0 2 5
## [26] 4 1 2 1 0 6 0 2 5 7 4 9 3 2 1 0 1 2 2 3 1 1 6 5 0
## [51] 1 0 0 0 0 4 9 0 1 4 1 1 1 5 3 2 2 4 2 1 2 1 0 2 3
## [76] 3 4 1 0 3# Betweenness
betweenness(el)[betweenness(el) >= 100]## [1] 603.0000 145.8333 112.0000 119.0000 154.0000# Closeness
closeness(nw)[closeness(nw) >= 0.008]## numeric(0)These nodes would be the most successful in spreading influence among the cycles that they are located in. Node 1 suffers from mental health issues while nodes 33 and 57 do not. Participants 33 and 57 also spend most of their time with either friends and family and do not have any direct ties with any individuals with mental health signs. Therefore, participants 33 and 57 would be great candidates to serve as mentors for those individuals struggling with their mental health. By including participant 1 in 33 and 57’s cycle, participant 1 can benefit from their positivity and in turn spread that line of support to the rest of 1’s cycle in hopes to create a support chain that benefits the network entirely.
Dimension Similarity Matrices
In order to set up the statistic tests, similarity matrices were built to describe the network with regard to certain characteristics. These similarity matrices were formed as adjacency matrices between the nodes of the network regarding a dimension of the "drugAddiction_Revised" dataset. If the value of one node for a dimension examined in the matrix was the same value as the dimension for another node, then a "1" would be put in the corresponding place in the adjacency matrix. For instance, if a person had a "1" value for the binary dimension of "Mental_problem" and another person also had a mental problem, a "1" would be marked in the similarity matrix. And naturally, if a person did not have a mental problem, a "0" would be placed in the similarity matrix. And so, the matrices would describe the overall characteristics of the matrix from the end of the characteristics described. As will be discussed in the next section, seven matrices were formed regarding seven dimensions of the "drugAddiction_Revised" dataset. Three of the dimensions regarded emotional, mental health-related factors, specifically whether a person has a mental problem ("Mental_problem"), whether they feel like a failure in life ("Failure_in_life"), and whether they have suicidal thoughts ("Suicidal_thoughts"). The four other dimensions regarded social factors, specifically the group of people a person spends of their time with ("Spend_most_time"), whether a person lives with family ("Live_with"), the quality of familial relationships ("Family_relationship"), and the number of friends a person has ("Number_of_friends"). The similarity matrices of thesse dimensions were compared in the next section with statistical tests.The code below creates the similarity matrices of the seven dimensions examined for the statistical tests. Each factor was processed through the “buildSimilarityMatrix” function to build the similarity adjacency matrix, comparing the dimension value for each node with each other node. The result of the function was a binary adjacency matrix with a “0” or “1,” comparing similarities in the nodes, or “0” for any node with itself.
# Build similarity adjacency matrix for dimension factor "dim"
buildSimilarityMatrix <- function(dim) {
sim_mat <- mat
column <- data.frame(dt[c(dim)])
for (i in 1:dim(sim_mat)[1]) {
for (j in 1:dim(sim_mat)[2]) {
if (i == j) {
sim_mat[i, j] <- 0
}
else if (column[i,1] == column[j,1]) {
sim_mat[i, j] <- 1
}
else {
sim_mat[i, j] <- 0
}
}
}
return(sim_mat)
}
# Build mental health factor similarity matrices
mental_problem_adj <- buildSimilarityMatrix("Mental_problem")
failure_in_life_adj <- buildSimilarityMatrix("Failure_in_life")
suicidal_thoughts_adj <- buildSimilarityMatrix("Suicidal_thoughts")
# Build social factor similarity matrices
live_with_adj <- buildSimilarityMatrix("Live_with")
spend_most_time_adj <- buildSimilarityMatrix("Spend_most_time")
family_relationship_adj <- buildSimilarityMatrix("Family_relationship")
number_of_friends_adj <- buildSimilarityMatrix("Number_of_friends")Statistical Tests
Statistical tests were drawn to examine the correlation of characteristics in the mental health network. The characteristics were drawn up as similarity matrices to describe the relationship between each node with regard to the characteristic involved. The statistical tests compared the similarity matrix of a mental health quality to that of the four social factors examined. The statistical tests were done using QAP linear regression to define the correlation between each a matrix and a comparable model. In this case, the mental health factor was set as the base adjacency matrix and then a model of the adjancency matrices of the four social factors were compared to analyze the correlation between each of the social factors with the original mental health characteristic. Overall, three statistical tests were done, comparing the social factors against whether a person had a mental problem, whether a person felt like a failure in life, and whether a person had suicidal thoughts. The social factors were attributed in the summaries of the linear regression models with "x"-denoted variables. The "x"-variables are described below.:- x1 — Whom A Person Lives With (Whether alone or with family)
- x2 — With Whom A Person Spends Most of Their Time (Whether alone, with friends, or with family)
- x3 — Family Relationship (Whether a communication gap, average, or satisfactory)
- x4 — Number of Friends (Whether none, few, or many)
The code below processes the mental_problem_adj, failure_in_life_adj, and suicidal_thoughts_adj matrices individually against the live_with_adj, spend_most_time_adj, family_relationship_adj, and number_of_friends_adj adjacency matrices. This was done using the "netlm" function of the SNA package for QAP linear regression, basing the code on Murphy (n.d.). The summaries of the three regression models are displayed below.cat("MENTAL PROBLEM CORRELATION MODEL:\n")## MENTAL PROBLEM CORRELATION MODEL:# Compare mental problem factor to social factors with QAP linear regression
nl<-netlm(mental_problem_adj,
list(live_with_adj, spend_most_time_adj, family_relationship_adj, number_of_friends_adj),reps=250)
# Show summary of linear regression model
summary(nl)##
## OLS Network Model
##
## Residuals:
## 0% 25% 50% 75% 100%
## -0.8210099 -0.6752402 0.2410919 0.2718658 0.3247598
##
## Coefficients:
## Estimate Pr(<=b) Pr(>=b) Pr(>=|b|)
## (intercept) 0.74352530 1.000 0.000 0.000
## x1 0.04671069 0.972 0.028 0.028
## x2 0.03077393 0.804 0.196 0.432
## x3 -0.05289400 0.116 0.884 0.232
## x4 -0.01539110 0.524 0.476 0.896
##
## Residual standard error: 0.4361 on 6315 degrees of freedom
## Multiple R-squared: 0.00786 Adjusted R-squared: 0.007231
## F-statistic: 12.51 on 4 and 6315 degrees of freedom, p-value: 3.902e-10
##
##
## Test Diagnostics:
##
## Null Hypothesis: qap
## Replications: 250
## Coefficient Distribution Summary:
##
## (intercept) x1 x2 x3 x4
## Min -11.03235 -2.18195 -5.81795 -8.05554 -10.07476
## 1stQ 2.92743 -1.08839 -2.65160 -3.47385 -7.39465
## Median 6.78440 -0.32168 -0.23623 -0.38992 -1.14448
## Mean 6.48870 0.19943 -0.02886 -0.09423 0.25149
## 3rdQ 10.37651 0.86069 2.25051 2.26553 5.52900
## Max 19.04491 9.98810 12.90126 14.75708 31.64618cat("\n--------------------------------\n")##
## --------------------------------cat("FAILURE IN LIFE CORRELATION MODEL:\n")## FAILURE IN LIFE CORRELATION MODEL:# Compare failure in life factor to social factors with QAP linear regression
nl2<-netlm(failure_in_life_adj,
list(live_with_adj, spend_most_time_adj, family_relationship_adj, number_of_friends_adj),reps=250)
# Show summary of linear regression model
summary(nl2)##
## OLS Network Model
##
## Residuals:
## 0% 25% 50% 75% 100%
## -0.5935329 -0.5006849 0.4263105 0.4823971 0.5480079
##
## Coefficients:
## Estimate Pr(<=b) Pr(>=b) Pr(>=|b|)
## (intercept) 0.47922925 1.000 0.000 0.000
## x1 0.01984338 0.908 0.092 0.092
## x2 0.02884939 0.916 0.084 0.084
## x3 0.06561085 0.992 0.008 0.008
## x4 -0.02723716 0.208 0.792 0.424
##
## Residual standard error: 0.4988 on 6315 degrees of freedom
## Multiple R-squared: 0.005412 Adjusted R-squared: 0.004782
## F-statistic: 8.591 on 4 and 6315 degrees of freedom, p-value: 6.541e-07
##
##
## Test Diagnostics:
##
## Null Hypothesis: qap
## Replications: 250
## Coefficient Distribution Summary:
##
## (intercept) x1 x2 x3 x4
## Min -4.10503 -1.46663 -1.90344 -2.21628 -3.83791
## 1stQ 2.22521 -0.84589 -1.02556 -1.23400 -1.33546
## Median 4.41305 -0.48214 -0.45876 -0.47406 -0.31133
## Mean 4.18848 0.14033 0.03126 -0.04751 0.24015
## 3rdQ 6.16487 0.39032 0.72783 0.64201 1.18589
## Max 12.60542 17.05968 6.74547 6.97661 8.04545cat("\n--------------------------------\n")##
## --------------------------------cat("SUICIDAL THOUGHTS CORRELATION MODEL:\n")## SUICIDAL THOUGHTS CORRELATION MODEL:# Compare suicidal thoughts factor to social factors with QAP linear regression
nl3<-netlm(suicidal_thoughts_adj,
list(live_with_adj, spend_most_time_adj, family_relationship_adj, number_of_friends_adj),reps=250) #gender_adj, live_with_adj,
# Show summary of linear regression model
summary(nl3)##
## OLS Network Model
##
## Residuals:
## 0% 25% 50% 75% 100%
## -0.7026277 -0.5940143 0.3044362 0.3987128 0.6022751
##
## Coefficients:
## Estimate Pr(<=b) Pr(>=b) Pr(>=|b|)
## (intercept) 0.404788824 1.000 0.000 0.000
## x1 -0.007063919 0.460 0.540 0.760
## x2 0.007272936 0.660 0.340 0.848
## x3 0.094276572 0.984 0.016 0.016
## x4 0.196289361 0.976 0.024 0.024
##
## Residual standard error: 0.4789 on 6315 degrees of freedom
## Multiple R-squared: 0.03867 Adjusted R-squared: 0.03806
## F-statistic: 63.51 on 4 and 6315 degrees of freedom, p-value: 0
##
##
## Test Diagnostics:
##
## Null Hypothesis: qap
## Replications: 250
## Coefficient Distribution Summary:
##
## (intercept) x1 x2 x3 x4
## Min -11.48402 -1.89052 -4.51016 -5.53544 -9.81407
## 1stQ 1.45944 -0.98159 -2.04545 -2.14797 -4.23500
## Median 4.40012 -0.48372 -0.61182 -0.42119 -0.77556
## Mean 3.87204 0.21892 -0.07550 -0.07715 -0.36901
## 3rdQ 7.09779 0.96234 1.26857 1.53589 3.22224
## Max 17.41921 8.02834 10.65638 12.49189 21.09793Analysis of Statistical Tests
The statistical tests brought insights into the most correlative factors to mental health factors. First, the model for whether someone has a mental problem was examined. Looking at the "Pr(>=|b|)" value for the four social factors, it appears "x1" was most correlative, followed by "x3," "x2," and "x4." Therefore, the matter of whether a person lived by themselves or with family was the most correlative factor to whether a person had a mental problem. This is followed by the factor of one familial relationships, whom one spends most of their time with, and the number of friends one has. It seems logical that the "Live_with" factor would be the most determining factor of the four. A person who lives themselves may be more apt to loneliness and depression than one who has a family supporting them. Inversely, it could also be that a person with who lives with their family has a greater likelihood of mental health problems due to stress and troubled relationships. No conclusion could be totally drawn, but it does give that whether one lives alone or with family is the most correlative factor to whether someone has a mental health problem. The other three make sense in turn with one's familial relationships having a good impact on one's mental health, then the people with whom one regularly associates, and one's number of friends.
The second model for whether someone felt like a failure in life drew presumably even closer ties with the social factors. The "Pr(>=|b|)" values for the "x"-values described were lower overall in the second model than the mental problem characteristic. The familial relationship characteristic was around 0.048, followed by the live with factor at around 0.088, the spend most time factor at around 0.108, and finally the number of friends at around 0.424. This generally makes sense as the first model because of the importance of familial relationships. It is interesting that familial relationships surpass the correlation of whom one lives with like the first model. It appears that one feeling like a failure in life draws from their family. This does make sense considering how one grows up, and the expectations one's family puts upon them. Of course, this could be correlative for better or for worse with regard to the familial connections of person. The cannot be fully concluded from this regression model. But, in this case, family has a far greater impact than friends according to the second model.
Finally, the third model for whether a person had suicidal thoughts contrasted a bit from the previous two models. The most correlative factor appeared to be "x4," the number of friends one has, followed by family relationships, and then quite a bit less, whom one lives and spends time with. The number of friends is interesting as it shows the impact of peers one life. The regression does state whether more friendship ties correlate to more suicidal thoughts or the inverse, but the quantity of friends does play significant part in one's suicidal tendencies. Perhaps this could be explained by if someone does not have a lot of friends, feels lonely, and finds little purpose. Alternatively, this could be interpreted as someone with a lot of friends compares themselves dramatically to others and maybe considers others friends but is not close with any individual people. Whatever the case, again there is a high correlation in mental health to familial connections. There is almost as high a correlation with the relationship one has with their family as with the number of friends. Therefore, this affirms the notion of the importance and impact one's family has on their life.
All three of the models above had a minimal p-value, being less than 0.05, giving that the models were useful (Murphy, n.d.). The residual standard error was also comparable and not overwhelming to indicate some of the validity of the correlation models. Overall, the three models all drew a tie familial connections had a significant correlation with one's mental health, being the foremost or second-most explanatory dimension of the social factors. The matter of with whom someone lived with was the next most important, followed by the number of friends and the people whom one spends the most time. It could be further analyzed how the specific values, say the quality of familial relationship, correlate to the mental health issues for better and for worse. Whatever the case, the impact of family and friendship relations was well-examined.Conclusion
The results of the survey among the participants helped us determine which attributes had the most influence on their mental well-being and help identify those actors who were most at risk to harm themselves. Centrality measures provided insight on which of the participants had the highest potential to increase dissemination throughout the network and incorporate them into cycles that can become foundations for strong and effective lines of support among the network. By connecting one high-risk individual to the participants that are more associated with other happy individuals, the foundation of a line of emotional support can established that would spread among the rest of the participants in need and increase the overall mental well-being of the rest of the network. In the statistical tests, it was further illustrated the factors that most correlative with mental health issues. The trend overall was that the quality of a person's relationship with their family had the overall most impact on whether someone had a mental problem, whether someone felt like a failure in life, and whether someone had suicidal tendencies. More research could be gathered on the further nature of the values of the social factors and the mental health issues for better or worse. In the end, the visual, centrality, and statistical analyses give guides to how to approach and diagnose problems and solutions for bettering people's mental well-being with regard to a social, interpersonal perspective.References
Atchley, R. C. (1991). The influence of aging and frailty on perception and expression of the self: Theoretical and methodological issues. The concept and measurement of quality of life in the frail elderly. New York: Elsevier.
Healy, K. (August 14, 2004). Love is a Spanning-Tree Network with no 4-Cycles. https://kieranhealy.org/blog/archives/2004/08/14/love-is-a-battlefield-spanning-tree-network-with-no-4-cycles/
kohske. (2020, January 16). How to unload a package without restarting R (Gaffi, MichaelChirico, R. Cotton, & A. B. Friedman, Eds.). https://stackoverflow.com/questions/6979917/how-to-unload-a-package-without-restarting-r
Murphy, P. (n.d.). Hypothesis testing in network analysis. https://rpubs.com/pjmurphy/338798
Protik, M.M.J. (2021). Drug Addiction in Bangladesh-Reasons. https://www.kaggle.com/protikmostafa/drug-addiction-in-bangladesh-reasons
Simpson, G. (2013, October 22). How to add new column to an dataframe (to the front not end)? https://stackoverflow.com/questions/19508256/how-to-add-new-column-to-an-dataframe-to-the-front-not-end
Zhou X., Liang X., Zhao J., et Zhang S. (August 6, 2018). Cycle Based Network Centrality. Retrieved from https://www.nature.com/articles/s41598-018-30249-4
Appendices
Not applicable.