Lonely Knight @stevenly83 - Tumblr Blog

Week 4 assignment - Machine Learning for Data

LIBNAME mydata "/courses/d1406ae5ba27fe300 " access=readonly;

************************************************************************************************************** DATA MANAGEMENT **************************************************************************************************************; data clust; set mydata.gapminder; run; * create a unique identifier to merge cluster assignment variable with the main data set;

idnum=_n_; keep idnum incomeperperson alcconsumption armforcesrate hivrate internetuserate urbanrate employrate suicideper100th lifeexpectancy femaleemployrate;

run;

ods graphics on;

* Split data randomly into test and training data; proc surveyselect data=clust out=traintest seed = 145 samprate=0.8 method=srs outall; run;

data clus_train; set traintest; if selected=1; run; data clus_test; set traintest; if selected=0; run; * standardize the clustering variables to have a mean of 0 and standard deviation of 1;

proc standard data=clus_train out=clustvar mean=0 std=1;

var incomeperperson alcconsumption hivrate internetuserate urbanrate employrate suicideper100th lifeexpectancy femaleemployrate; run; %macro kmean(K);

proc fastclus data=clustvar out=outdata&K. outstat=cluststat&K. maxclusters= &K. maxiter=30000;

var incomeperperson alcconsumption hivrate internetuserate urbanrate employrate suicideper100th lifeexpectancy femaleemployrate; run;

%mend;

%kmean(1); %kmean(2); %kmean(3); %kmean(4); %kmean(5); %kmean(6); %kmean(7); %kmean(8); %kmean(9); * extract r-square values from each cluster solution and then merge them to plot elbow curve; data clus1; set cluststat1; nclust=1;

if _type_='RSQ';

keep nclust over_all; run;

data clus2; set cluststat2; nclust=2;

if _type_='RSQ';

keep nclust over_all; run;

data clus3; set cluststat3; nclust=3;

if _type_='RSQ';

keep nclust over_all; run;

data clus4; set cluststat4; nclust=4;

if _type_='RSQ';

keep nclust over_all; run; data clus5; set cluststat5; nclust=5;

if _type_='RSQ';

keep nclust over_all; run; data clus6; set cluststat6; nclust=6;

if _type_='RSQ';

keep nclust over_all; run; data clus7; set cluststat7; nclust=7;

if _type_='RSQ';

keep nclust over_all; run; data clus8; set cluststat8; nclust=8;

if _type_='RSQ';

keep nclust over_all; run; data clus9; set cluststat9; nclust=9;

if _type_='RSQ';

keep nclust over_all; run;

data clusrsquare; set clus1 clus2 clus3 clus4 clus5 clus6 clus7 clus8 clus9; run;

* plot elbow curve using r-square values; symbol1 color=blue interpol=join; proc gplot data=clusrsquare; plot over_all*nclust; run;

A k-means cluster analysis was conducted to identify underlying subgroups of people's income based on their similarity of responses on 9 variables that represent characteristics that could have an impact on people's daily life in different countries. Clustering variables included quantitative variables measuring alcohol consumption, a scales measuring HIV rate, internet use rate, life expectancy, urban rate, employ rate, people who suicide and female employ rate. All clustering variables were standardized to have a mean of 0 and the standard deviation of 1.

Data were randomly split into a training set that included 80% of the observations (N= 171) and a test set that included 20% of the observations.A series of k-means cluster analyses were conducted on the training data specifying k=1-9 clusters, using Euclidean distance. The variance in the clustering variables that was accounted for by the clusters (r-square) was plotted for each of the nine cluster solutions in an elbow curve to provide guidance for choosing the number of clusters to interpret.

In the figure below, the plot shows the increase in the proportion advantage in the clustering variables, explained by each of the cluster solutions. The two cluster solution accounts for about 28% or 29% of the variance. The R-square value increases as more clusters are specified

Machine Learning for Data Analysis- Assignment week 3

Below is the code used for executing this analysis.

data new; set mydata.treeaddhealth; if bio_sex=1 then male=1; if bio_sex=2 then male=0;

* delete observations with missing data; if cmiss(of _all_) then delete; run;

ods graphics on;

* Split data randomly into test and training data; proc surveyselect data=new out=traintest seed = 150 samprate=0.8 method=srs outall; run;

* lasso multiple regression with lars algorithm k=10 fold validation; proc glmselect data=traintest plots=all seed=150; partition ROLE=selected(train='1' test='0'); model famconct = male hispanic white black namerican asian alcevr1 marever1 cocever1 inhever1 cigavail passist expel1 age alcprobs1 deviant1 viol1 dep1 esteem1 parpres paractv schconn1 gpa1/selection=lar(choose=cv stop=none) cvmethod=random(10); run;

A lasso regression analysis was conducted to identify a subset of variables from a pool of 23 categorical and quantitative predictor variables that best predicted a quantitative response variable measuring family connectedness in adolescents. Categorical predictors included gender and a series of 5 binary categorical variables for race and ethnicity (Hispanic, White, Black, Native American and Asian) to improve interpretability of the selected model with fewer predictors.

Similar with the one that we learned in class, binary substance use variables were measured with individual questions about whether the adolescent had ever used alcohol, marijuana, cocaine or inhalants. Additional categorical variables included the availability of cigarettes in the home, whether or not either parent was on public assistance, and any experience with being expelled from school. Quantitative predictor variables include age, alcohol problems, and a measure of deviance that included such behaviors as vandalism, other property damage, lying, stealing, running away, driving without permission, selling drugs, and skipping school. Another scale for violence, one for depression, and others measuring self-esteem, parental presence, parental activities, school connectedness and grade point average were also included. All predictor variables were standardized to have a mean of zero and a standard deviation of one.

Data were randomly split into a training set that included 80% of the observations (N=4572) and a test set that included 20% of the observations (N=914). The least angle regression algorithm with k=10 fold cross validation was used to estimate the lasso regression model in the training set, and the model was validated using the test set.

The number of parameters to be estimated is 24 for the intercept plus the 23 predictors.

According to the table with the LAR selection information, it shows the steps in the analysis and the variable that in entered at each step. ASE and Test ASE are the means square error for the training data and the test data. They are ordered in terms of how important they are in predicting family connectedness. It appears that the most important predictor of family connectedness was self esteem followed by deviant behavior and so on. In this table, the average square error also declines as variables are added to the model, indicating that the prediction accuracy improves as each variable is added to the model.

Machine Learning for Data Analysis - Week 2

Today, we'll continue defining the target variable based on answers to the question, have you ever drunken alcohol? The candidate explanatory variables include gender, race, marijuana, cocaine or inhalant use, availability of cigarettes in the home, whether or not either parent was on public assistance, any experience with being expelled from school.

LIBNAME mydata "/courses/d1406ae5ba27fe300 " access=readonly;

DATA new; set mydata.treeaddhealth;

PROC SORT; BY AID;

PROC HPFOREST;

target ALCEVR1/level=nominal;

input BIO_SEX HISPANIC WHITE BLACK NAMERICAN ASIAN MARever1 cocever1 inhever1

Cigavail PASSIST EXPEL1 /level=nominal;

input age DEVIANT1 VIOL1 DEP1 ESTEEM1 PARPRES PARACTV

FAMCONCT schconn1 GPA1 /level=interval;

run;

After I ran the program, the output below shows that the program ran locally and that two threads were used. No parameters are specified in the PROC HPFOREST statement; therefore, all the values are default. The variables to try is equal to 5, indicating that a random selection of 5 explanatory variables was selected to test each possible split for each node in each tree within the forest. By default, SAS will grow 100 trees and select 60% of their sample when performing the bagging process. That is the inbag fraction. The prune fraction specifies the fraction of training observations that are available for pruning a split.

The next figure shows that the number of observations read from data set was 6,504 while the number of observations used was 6,444. It also shows the baseline fit statistics. The listing shows a baseline misclassification rate of 0.448, which is 44.8% of the sample. It's suggesting that the forest correctly classified 55.2% of the sample.

As you can see the first 10 and the last 10 observations of the fit statistics. When PROC HPFOREST runs in single machine mode, it computes fit statistics for a sequence of forests that have an increasing number of trees. As the number of trees increases, the fit statistics usually improve at first and then level off and fluctuate in a small range. The out of bag misclassification rate for the model fluctuates between 0.279 and 0.285, which is in a range much less than the baseline rate of rate of 0.448. Therefore, you can conclude that the model is good.

Machine Learning for Data Analysis- Week 1 Assignment

It has been awhile that I write blog again for my school assignment.

For this week's assignment, it involves decision trees that are predictive models that allow for a data driven exploration of nonlinear relationships and interactions among many explanatory variables in predicting a response or target variable. In this assignment, I choose 15592 as the number seed which is used in the cross validation process while including PROC HPSPLIT to build tree based statistical models for classification regression.

In the model information, the decision tree that SAS grew has 143 leaves before pruning and 7 leaves following pruning. The target here is ALCEVR1, for alcohol drinker. In this assignment, I want to continue with my research on alcohol dependence. However, my model event level is predicting the value 0, that is no for our target variable, alcohol ever.

The number of observations read from my data set was 6564 while the number of observations used was only 4575. 4575 represents the number of observations with valid data for my target variable and each of the explanatory variables. Those observations with missing data on even one variable have been set aside.

For the next plot of the cross validated average misclassification rate (AMR), instead of ASE, average standard error. Some how the result came out a little bit difference than I expected even though I used the same program and tree add health as the lesson. A vertical reference line is drawn for the tree with the number of leaves that has the lowest cross validated AMR. In this case, there is only 5 leaf tree.

Following the pruning plot that chose a general model with 9 split levels and 7 leaves, the final, smaller tree is presented, which shows the model with splits on alcohol problem, marijuana use, and deviant behavior.

Moreover, based on the model-based confusion matrix, it shows how well the final classification tree performed. The total model correctly classifies 83% of those who have not ever drunken alcohol. That is 1 - the error rate of .17 and 70% of those who have. It helps us predict those who are protected against alcohol drinker during adolescence, and less likely to predict those who are at risk for alcohol drinking.

Finally SAS provides a variable importance table. The importance score measures a variable’s ability to mimic the chosen tree and to play the role as stand-in for variables appearing as primary splits. Here we see that variables such as marijuana ever and deviant behavior, have important scores that are relatively similar to alcohol problems which was selected as a split in our final model.

Regression Modeling in Practice -Assignment 4

Even though my response variable is not binary yes or no like nicotine dependence, I made it into 3 separate binary response variables such as alcohol abuse, alcohol dependence and alcohol abuse and dependence when I work on Logistic Regression for a Binary Responsible variable. We also have an explanatory variable called MAJORDEPLIFE that indicates the presence or absence of major depression. In this assignment, I want to see how major depression relates with people who abuse or dependence on alcohol or both.

In this output, I can see the number of observations and the number of observations with complete data that were used in the model.

However, I notice that my regression is not significant at an alpha level of 0.1159 which is greater than .005 when I look at the parameter estimates and the p value.

As I move to the next output of alcohol dependence, I also notice the regression is not significant as p is greater than 0.005.

Only our last regression is significant at an alpha level less than 0.0001. Using the parameter estimates, we could also generate the linear equation: alcoabusedep = -1.8 + 1.3(MAJORDEPLIFE).

Moreover, as you can see, based on my odds ratio, it is greater than 1. It means that the probability of becoming alcohol abuse and dependent increases among those with major depression compared to those without. In another way, those adult who drink alcohol daily with major depression are 3.7 times more likely to have alcohol abuse and dependence than adult who drink alcohol without major depression.

In additional, I also get a confidence interval for my odds ratio. This confidence interval tells us that we can be 95% confident that if I select another sample from the population, the odds ratio for that new sample will be somewhere between 2.56 and 5.27 95 times out of 100.

The relationship between alcohol consumption and earnings

Last week, in our last discussion, we know that there is a statistically significant association between income and alcohol use. The study found people with lower incomes had more variation in how much they drank, compared with people with higher incomes.

It appears that the low income group includes more light drinkers and non-drinkers, as well as more heavy drinkers, than the high inocme group. People with higher incomes, in contrast, are more likely to drink overall, but they are also more likely to moderate how much alcohol they consume.

Going back to the SAS program, we first need to find the mean of income per person who consumes alcohol by using the mean procedure. The mean is 8740.97

We then add this code to center the number of income per person variable using the means procedure.

data new2;

set new;

incomeperperson_c= incomeperperson - 8740.97;

proc means;

var incomeperperson_c;

run;

The mean here is -0.004 which is almost centers the variable at zero. We now can use the centered variable in our regression analysis.

Let’s test a regression model for just a linear association between alcohol consumption and income per person using the GLM procedure.

According to the result of the significant P value and positive parameter estimate which is very close to 0, we can conclude that the alcohol consumption is positively associated with income per person but very slightly. The R square is close to 0.09 indicating that the linear association of income per person is capturing only about 9% of the variability in alcohol consumption.

*polynomial regression model; PROC GLM; model alcconsumption = incomeperperson_c incomeperperson_c*incomeperperson_c/solution clparm; run;

Now, when we look at the table result of polynomial after we square the income per person_c variable, the value of the linear term for incomer per person is positive and the p value is less than 0.05. However, the quadratic term is negative, and significant indicating that the curvilinear pattern we observed in our scatter plot is statistically significant.

A positive linear coefficient and a negative quadratic coefficient indicates that the curve is concave downward.

In addition, we see that the R- Square increased at 0.12, which means that adding the quadratic term for income per person increased the amount of variability in alcohol consumption.

PROC GLM PLOTS(unpack)= all; model alcconsumption = incomeperperson_c incomeperperson_c*incomeperperson_c employrate_c/ solution clparm; output residual=res student=stdres out=results; run;

Now, we using glm procedure to test regression model. The intercept is the value of the response variable when all the explanatory variables are held constant at a value of 0. Like the lesson that I learned this week, I centered two explanatory variables, so that the mean for each variable was equal to 0, the intercept is the alcohol consumption at the mean of income per person and employ rate. So, the alcohol consumption rate when income per person and employ rate are at their mean is 8 out of every 100 people. The coefficients for the linear and quadratic of income per person variables remain significant after adjusting for employ rate. However, income per person and employ rate together explain only about 12% of the variability in alcohol consumption. It means there is some error in estimating the response value with this model.

Through Q-Q plot, you can see that the residuals kind of follows a straight line, but deviate somewhat at the lower and higher quantiles. This indicates that our residuals do not follow perfect normal distribution.

When we look at the second residual plot for each observation at different values of income per person, it looks like a funnel shaped pattern to the residuals since most of the values of the residuals are significant larger at lower values of income per person and get smaller that is closer to 0 as income per person increases.

For the first residual plot, I can’t find the explanation for it yet. Hope to hear your opinion about it at the comment.

Week 2 Assignment program

libname mydata "/courses/d1406ae5ba27fe300" access=readonly;

data new; set mydata.gapminder; run;

********************************************************************************************* BASIC LINEAR REGRESSION ****************************************************************************************************;

* scatterplot with linear regression line; PROC sgplot; reg x=incomeperperson y=alcconsumption / lineattrs=(color=blue thickness=2); title "Scatterplot for the Association Between Income and Alcohol Consumption"; yaxis label="Alcohol Consumption"; xaxis label="Income"; run; title;

* basic linear regression; PROC glm; model incomeperperson=alcconsumption/solution; run;

The association between alcohol and income

Week 2 Assignment- Regression Modeling in Practice

Excessive alcohol use poses an important public health problem. Approximately 2 billion people consume alcohol worldwide. Alcohol abuse accounts for approximately 105,000 deaths each year in the United Staets, due to intentional and unintentional injury, cirhosis, stroke, pneumonia and cancers of the digestive system. Almost two-fifths of traffic fatalities are the result of alcohol abuse.

Income may be an important driver of alcohol use. Income directly measures access to material resources and to health-enhancing environments and services, as well as exposure to material stressors such as financial problems and deprivation.

Prior research has show that while lower income individuals are at higher risk of engaging in heavy, hazardous drinking and abstinence, higher income is associated with a higher frequency of light drinking.

If heavy drinking among lower-income individuals use alcohol as a way to relieve stressful life experiences, frequent light consumptions among individuals of higher income may be related to the availability of disposable income to purchase alcohol, social norms supportive of light, frequent drinking in middle and high income social groups, and social drinking in the context of social and job-related networking.

In this week assignment, by using GapMinder Codebook, and observing 179 people with different income level, we now know there’s a statistically significant association between income and alcohol use in order to support the statement that we mentioned above.

From our analysis, the F statistic is 15.92 and the P value is very small considerably less than our alpha level of .05, which tells us that we can reject the null hypothesis and conclude the income per person is significant associated with the alcohol consumption.

Moreover, when we look at the parameter estimates at the end of our output, the data sub one bay here is 610.06 and the beta sub zero value is 2891.32. So we now know that our equation for the best fit line of this graph is

incomeperperson = 2891.32 + 610.06 * alcconsumption.

By looking at how the equation was written, you can see that even with 15 liters of alcohol consumptions as heavy drinking, the income of a person is still under low-income at $12,042.

Also, if we look at the output for the coefficients, we have a column labled Pr greater than the absolute value of t, which gives us the p value, that is very small for our explanatory variables association with their response variable. The GLM Procedure also gives an R-Square value is the proportion of variance in the response variable that can be explained by the explanatory variable. About 8.7% or 9% of the variability we see in our response variable income per person.

Moreover, by looking at scatter plot, you could see that the group of people who consume the most alcohol beverage, are the people who have low- income under $20,000. The lower the income, there are more people consume the large amount liters of alcohol.

Assignment 1- Writing About My Data

Sample

The sample is from the first wave of the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC), the largest nationwide longitudinal survey of alcohol and drug use and associated psychiatric and medical comorbidities. Participants (N=43,093) represented the civilian, non-institutionalized adult population of the United States, and included persons living in households, military personnel living off base, and persons residing in the following group quarters: boarding or rooming houses, non-transient hotels and motels, shelters, facilities for housing workers, college quarters, and group homes. The data analytic sample for this study included participants who ever felt sad, blue, depressed or down most of the time in 2 week period (N=12785).

Procedure

Data were collected by trained U.S. Census Bureau Field Representatives during 2001– 2002 through computer-assisted personal interviews (CAPI). One adult was selected for interview in each household, and interviews were conducted in respondents’ homes following informed consent procedures.

Measures

A series of yes no questions have been asked by the researcher in order to find out the detailed symptoms, conditions that might occur on a person who has lifetime major depression. It helps us recognize the link of all the symptoms in the first 2 weeks period to know if a person has major depression or not. It’s because of the existence of a virtually unlimited number of potential lurking variables that we can never be 100% certain of the claim of causation based on an observational study. Based from what we learned, observational studies can’t prove causality.

Data Analysis Tool- Assignment 4

Hello everyone;

Thank you for spending your time to read my blog. I hope you all had a great time with your family so far on Christmas day and enjoying the long weekend before we welcome a New Year.

In the last assignment of this course, for the association between alcohol assumption and life expectancy, the correlation coefficient is approximately 0.30 with a p-value of 0.0001. This tells us that the relationship is statistically significant.

Similar with the testing moderation in the context of correlation of lesson 4, does income level moderate the relationship between alcohol assumption and life expectancy?

We sort the income data by new categorical third variable, which we called income group. It was categorized as a high income country given a value of 3, a moderate income country given a value of 2 and a low income country given a value of 1.

Then when we examine the correlation coefficients between the alcohol assumption and life expectancy for each of the income groups, we find the following.

For the low income group, the correlation between alcohol assumption and life expectancy is negative 0.11 and the p value is not significant.

For the moderate income countries, the correlation between alcohol assumption and life expectancy is 0.12 with also a large p-value. So far we could suggest that the association between alcohol assumption and life expectancy are not significant for low income and moderate income countries.

Finally, among high income countries, the correlation coefficient is 0.44 and the p value is 0.0051. From what I studied, the p value has to be less than 0.005 in order to be considered significant. In this case, the p value is slightly higher than 0.005, then it’s also not significant

When we map these findings onto the associated scatter plots for each income group, we are able to better visualize the non-significant relationships. Different from income group 2 and 3, it shows the negative association between alcohol assumption and life expectancy in low income countries.

Data Analysis Tool- Week 3 Assignment

This week, we learned about how to calculate correlation in SAS Program. It’s the most simplest one that I learned so far about SAS Program.

This week, we are going to be looking at 2 different correlations, that between alcohol assumption and suicide, and between alcohol assumption and income per person.

Together, let look at the outcome below

According to the Pearson Correlation Coefficient table, for the association between alcohol assumption and suicide, the correlation coefficient is approximately 0.35 with a p-value of 0.0001. This tells us that the relationship is statistically significant.

For the association between income per person and alcohol assumption, the correlation coefficient is approximately 0.29 and also has a significant p- value.

However, both the association between alcohol assumption and suicide and the association between alcohol assumption and income per person are weak even though they are positive as the scatter plot.

Beside, if we square the correlation coefficient, we also get value that helps our understanding of the association between the two quantitative variables.

When looking at the relationship between alcohol assumption and suicide, if we square our correlation coefficient of 0.35, we get 0.12. It means that if we know the “ alcohol assumption,” we can predict only 12% of the variability we will see in the rate of suicide. Of course, that also means that 88% of the variability is unaccounted for.

Same thing, if we square the correlation coefficient for income per person and alcohol assumption, we get a value of 0.084. This suggests, if we know income per person, we can predict 8% of the variability we will see in the rate of alcohol assumption.

Data Analysis Tool- Week 2 Assignment

. This week, I am going to use the Chi-Square test of independence to test the hypothesis about using cocaine frequency and major depression from working with NESARC data. In another words, I want to see is how often a person uses cocaine related to major depression among young adults? In this analysis, I am going to use a categorical explanatory variable with 9 levels. The number of days used cocaine per year, which in the program, I called USFREQYR with the following categorical values. They are 1 day per year, 2 days per year, 4.5 days per year, 9 days per year, 12 days per year, 30 days per year, 72 days per year, 168 days per year and 365 days per year. My response variable is categorical with 2 levels, that is the presence or absence of major depression in people’s life called MAJORDEPLIFE in the NESARC data set.

Since I am interested in describing the presence of depression within the using cocaine frequency groups. My p-value for the chi square test of 0.5 is clearly assuming that the two variables are not related. There is no compelling statistical evidence to reject the null hypothesis. Even if I use SAS code to graph the percent of young adults with major depression with each using cocaine frequency category, I could see that there is no association relationship between those two variables. The more days per year a young adult uses cocaine, it’s not likely there are presence of depression.

Moreover, this week, I also learned that when the explanatory variable has more than 2 levels, the Chi-Square statistic and associated p value do not provide insight into why the null hypothesis can be rejected. To determine which groups are different from the others, I perform a post hoc test on the same NESARC data that I researched last week by conducting post hoc comparisons between pairs of rates in a way that avoids excessive type 1 error. When we observe the frequency of how often people use cocaine or cracking in the last 12 months, we could see that there are 9 groups total as I combine “every day” and “nearly everyday” in 1 group. Therefore, we actually need to perform 36 pairwise comparisons.

In this week lesson, in order to protect against type 1 error in the context of the Chi-Square test, we will use the post hoc approach known as the Bonferroni Adjustment. The adjusted p value is calculated by dividing p .05 by the number of comparisons that we plan to make.

Since there are 36 number of comparisons, the adjusted p value is .05/36 = .00139.

Like the lesson that we learned, the goal is to examine the p value for each of the paired comparisons and to use the adjusted Bonferroni p value of .00139 to evaluate significance.

As you can see, I chose 2 using cocaine frequency groups at a time to compare one by one. Again, I wanna focus on the column percentages to see if these two rates significantly different from one another.

By looking down at my Chi-Square value and probability, I can see that they aren’t in most of the cases. The probability value of most cases are not only not less than 0.05, it is definitely not less than my Bonferroni Adjusted p value of .00139.

Data Analysis Tool- Week 1 Assignment

In my first assignment of this week, I want to do a research to find out if major depression is associated with using cocaine or crack among current young adult from the age of 18 to 40. Similar with the research that we learned in class, I want to look at the mean number of grams per year equal or not equal for those individuals with and without major depression?

The explanatory variable here is the presence or absence of major depression while the response variable is cocaine or crack quantity, which is measured the number of grams of cocaine or crack using per year.

It was shock to me to see the result of the ANOVA as I run SAS program and follow all the steps that I learned in class.

Even though I spread out the age of research from 18 to 40 years old, out of the 42 observations, 9 were included in the analysis

According to ANOVA, our calculated F statistic, or the F value in this output is 2.21. The significance, probability or p value associated with this F statistic, is labeled Pr > F. And as you can see, the P value is 0.1808 which is over our P value .05 cut point.

If we look at the mean table below, we can see that young adults who use cocaine or crack without major depression, as indicated by a value of 0, use 12.857 grams of cocaine per year. And those with major depression, indicated by a value of 1, use on average of 1 grams of cocaine per year.

Because the P value is greater than 0.05, we must accept the null hypothesis and say that these mean the data do not provide enough evidence to reject the null hypothesis. In another words, there is no association between the presence or absence of major depression and the number of grams of cocaine or crack use by young adults.

Data Management & Visualization- Assignment Week 4

Everyone feels sad or low sometimes, but these feelings usually pass with a little time. To be diagnosed with depression, symptoms must be present most of the day, nearly every day for at least 2 weeks. Moreover, depression affects different people in different ways. It can begin at any age. For example:

Older adults with depression may have less obvious symptoms, or they may be less likely to admit to feelings of sadness or grief. They are also more likely to have medical conditions, such as heart disease, which may cause or contribute to depression.

Younger children with depression may pretend to be sick, refuse to go to school, cling to a parent, or worry that a parent may die.

Older children and teens with depression may get into trouble at school, sulk, and be irritable. Teens with depression may have symptoms of other disorders, such as anxiety, eating disorders, or substance abuse

Generally, depression is split into two distinct categories: Early-Onset and Late-Onset.

Early-onset means the illness of depression is observed in childhood, adolescence or young adulthood.

Late-onset typically occurs in late adulthood, around 65 years of.

On the table and graph below, I divided the age of people who was onset of first episode into 4 groups.

Group 1 is the group of younger children from age of 5 to 10

Group 2 is the group of older children and teens from age of 11 to 19

Group 3 is the group of young adult below 40 years old

And group 4 is the group of older adult over 40 years old

Based on the graph below, we could see that group 1 and 3 have highest percentage of people who have first episode of depression.

Data Management class- Assignment week 3

In accord with DSM-IV, the first major depressive episode was defined by the first experience of a cluster of symptoms and signs of a major depressive episode, lasted at least 2 weeks, and was associated with clinically significant distress or functional impairment. Last week, we have a chance to compare the variables and find out how depression could affect your appetites and your weight. Today, I want to have a clearer picture about those sign and symptom of depression in a smaller group. To obtain clinically meaningful findings, the age-at-onset subgroups were defined in relation to the human life cycle. It has been suggested that younger patients tend to report lower ages at onset than older patients, but this has been contested.

Based on what I learned from last week and this week, I was able to set aside missing data in order to see a clear counts and percentage on each categories in early adult onset group from age 18 to 44.

Because we limit the number of the participants in a smaller group of people from age 18 to 44, there are 5112 participants in the total of 12785 participants.

In each categories that I work on, I don’t want to include the unknown response in my analysis. Therefore, I set the response to missing to ensure that SASS disregards these values. As you can see from my output, these variables have been managed in each categories comparing to before.

Moreover, according to the output, it was not surprising to me that people who are at the age 18 and 25 have higher frequency of first episode in depression than other ages in the group.

When you are at the age 18, you are in the border of teenager and young adult. You don’t know who you are and what is your future. You are being pressured by parents or people around you on choosing good college to go to or good career, good major to go with.

Then when you are at the age 25, you might just graduate from college and be struggling finding a good stable job with the degree that you have. This is also the age when you start to have a small family of your and then struggling on paying bill, or how to become a good parent, etc.

DATA new; set mydata.nesarc_pds; LABEL S4AQ1="Felt sad, blue, depressed in 2-week period" S4AQ4A1="Lost 2+ pounds per week for several weeks" S4AQ4A2="Lost appetite nearly everday for 2 weeks" S4AQ4A3="Gained 2+ pounds per week for several weeks" S4AQ4A4="Wanted to eat more than usual most days for 2 weeks" S4AQ6A= "AGE AT ONSET OF FIRST EPISODE";

/*comparing the data between losing weight and gaining weight because of depression in early adult onset from 18 to 44*/ IF S4AQ1 =1; IF S4AQ4A1=9 THEN S4AQ4A1=.; IF S4AQ4A2=9 THEN S4AQ4A2=.; IF S4AQ4A3=9 THEN S4AQ4A3=.; IF S4AQ4A4=9 THEN S4AQ4A4=.; IF S4AQ1 NE 9 AND S4AQ4A1=. THEN S4AQ4A1= 11; ELSE IF S4AQ1 NE 9 AND S4AQ4A2=. THEN S4AQ4A2= 11; ELSE IF S4AQ1 NE 9 AND S4AQ4A3=. THEN S4AQ4A3= 11; ELSE IF S4AQ1 NE 9 AND S4AQ4A4=. THEN S4AQ4A4=11; IF S4AQ6A LE 44; IF S4AQ6A GE 18;

PROC SORT; by IDNUM; PROC FREQ; TABLES S4AQ1 S4AQ4A1 S4AQ4A2 S4AQ4A3 S4AQ4A4 S4AQ6A; RUN;

Data Management and Visualization- Assignment of Week 2

Today, in this assignment, I want to continue my discussion in major depression. Different with last week, I want to focus on how depression could affect on our eating and weight in 2 week period by using the data managed variables provided from NESARC codebook, as frequency distributions. As I learn from NESARC codebook, 1 means yes, 2 means no and 9 means unknown.

In the survey of 12,785 people who felt sad, blue or depressed in 2 week period, I could compare and see that there are more people lost appetite and 2 pounds of weight because of depression than people who wanted to eat more and gained weight.

LIBNAME mydata "/courses/d1406ae5ba27fe300" access= readonly; DATA new; set mydata.nesarc_pds; LABEL S4AQ1=" Felt sad, blue, depressed in 2-week period" S4AQ4A1="Lost 2+ pounds per week for several weeks" S4AQ4A2="Lost appetite nearly everday for 2 weeks" S4AQ4A3="Gained 2+ pounds per week for several weeks" S4AQ4A4="Wanted to eat more than usual most days for 2 weeks"; /*comparing the data between losing weight and gaining weight because of depression*/ IF S4AQ1=1;

PROC SORT; by IDNUM; PROC FREQ; TABLES S4AQ1 S4AQ4A3 S4AQ4A1 S4AQ4A2 S4AQ4A4; RUN;

The Output

How does depression affect your life?

What is Depression? It is a common and serious medical illness that negatively affects how you feel, the way you think and how you act. After looking through the codebook for the NESARC study, I have decided that I am particularly interested in major depression. In the beginning, I am not sure which variables I will use regarding the depression, so I gather and include all of the relevant variables in my personal codebook. While major depression is a good starting point, I need to determine what it is about depression that I am interested in. Then it strikes me that people experience depression in different ways. It may interfere with a person's everyday activities, resulting in lost time and lower productivity. I used to be suffering from depression myself for many years. Therefore, I decide that I am most interested in exploring the association between level of how people experience depression in any given 2-week period and health conditions. Let first talk about sleep and depression. It has been estimated that 90% of patients with depression complain about sleep quality. Of all the psychiatric disorders associated with insomnia, depression is the most common one. Numerous sleep studies have provided extensive observations and theoretical hypotheses concerning the etiology and pathophysiology of depression. According to NESARC study, there are 7422 people had trouble falling asleep nearly everyday while there are 5783 woke up too early, usually in the middle of the night and couldn't go back to sleep nearly every day for the given 2 week period.

In both scenarios, we could assume that with people who lack of sleep or bad sleep quality, they will start to feel tired easily, increase fatigue, have trouble concentrating and lose appetite because they became so restless that felt uncomfortable. Of course, the common result of all of that is losing weight at least 2 pounds in 1 week.

However, not everyone who experiences depression will have insomnia or bad sleep quality. In contrast, they sleep more than usual nearly everyday since they don’t care about things around them or they want to spend more time than usual alone just because they don’t want to be around people.

In this scenario, it will not surprised for you to see people wanted to eat more than usual, and because of that, they also gain weight at least 2 pounds in 1 week.

According to NESARC study, there are only 2746 people gain weight for several weeks because of depression, while there are double the number of people lose same amount of weight for several weeks. I think that’s the reason why when we hear about depression’s symptom, it can vary from mild to severe and can include:

Feeling sad or having a depressed mood

Loss of interest or pleasure in activities once enjoyed

Changes in appetite — weight loss or gain unrelated to dieting

Trouble sleeping or sleeping too much

Loss of energy or increased fatigue

Increase in purposeless physical activity (e.g., inability to sit still, pacing, handwringing) or slowed movements or speech (these actions must be severe enough to be observable by others)

Feeling worthless or guilty

Difficulty thinking, concentrating or making decisions

Thoughts of death or suicide

Symptoms must last at least two weeks and must represent a change in your previous level of functioning for a diagnosis of depression.

Trending Blogs

Recently Viewed Blogs

Lonely Knight