Depression is a mental illness that leads to emotional distress and lethargy. Such an illness is often unbearable and disrupts a person's life making it difficult to enjoy life. Low mental health can be caused by a myriad of factors both genetic and environmental. Medication and therapy are used to treat depression. However, adopting certain behaviors and habits can better help to manage mental distress. My project is to analyze certain factors related to financial wellness to see if it produces an effect on the mental health of a person
My hypothesis is that lower economic status can make you more depressed. After all, not being able to afford the quality of life you would like to live should in theory make you more unhappy. I wanted to see how many services New York offers to help its citizens with both their financial health and their mental health.
According to NAMI, the national alliance on mental illness, 1 in 5 U.S. adults experience mental illness each year. Mental distress can negatively impact the quality of one’s life, making it difficult to function. Some people no longer feel the drive to go about their lives, unable to support themselves in education and employment. Others feel so despondent that they even take their lives. If we can narrow down factors that cause depression, we can do more to decrease its occurrence. I think one strong factor is one's economic status. While having more money might not instantly make you happier, it provides some stability and reduces problems related to money. Using features from open New York City datasets, I plan to analyze the trends in economic status and mental health. My prediction is a strong positive linear relationship between economic status and mental health. Lower economic status should correlate with lower mental health and higher economic status with higher mental status.
Economic status is the level of one’s financial health, like the ability to afford necessities like food and shelter, and also their wants. Economic status will be observed through a few features isolated in my Community Health Survey Dataset. These features are poverty group, neighborhood poverty group, inability to afford medication, employment level, and missing rent. More Information on these features is provided in my Data section.
Mental health status is one’s condition regarding psychological and emotional well-being. Mental Health status features are isolated in the Community Health Survey and include a mood rating, occurrence of psychological distress, and mental health treatment. People who experience some mental distress and are in treatment are considered to have a low mental health status. More Information on these features is provided in my Data section.
The Department of Health and Mental Hygiene conducts an annual telephone survey to collect various information about the health of New York citizens. Almost 10,000 New Yorkers from all kinds of backgrounds answered the survey. The features we chose to isolate were related to one’s mental health, financial health, and also demographic information.
MENTAL HEALTH FEATURES:
'Mood1': Have you ever felt so sad that nothing could cheer you up? Scale of 1 from all the time to 5 for none of the time.
'Nspd': Nonspecific Psychological Distress in the last 30 days, 1 for yes, 2 for no
'Mhtreat20_all': In Mental Health Treatment either therapy or taking medication. 1 for yes, 2 for no.
FINANCIAL HEALTH FEATURES:
'Imputed_povertygroup': Poverty Group ranging from below 100% poverty level to over 600%
'Imputed_neighpovgroup4_1519': Neighborhood Poverty groups ranging from low to very high
'Skiprxcost': Skipping medication because of inability to afford it in the past 12 months. 1 for Yes, 2 for no.
'Emp3': Employment Level which is Employed, Unemployed, and Not in WorkForce
'Delaypayrent': Missed rent in the past 12 months. 1 for yes, 2 for no.
DEMOGRAPHIC:
'Birthsex: Sex assigned at birth, 1 for Male, 2 for Female
'Agegroup': Age group ranging from 18-24 to 65 and up
Data2go compiles free open-source NYC data to use in mapping and visualization. I used the location data to make my choropleth visualizations on the number of psychiatric hospitalizations by community district and also the mean income level by city district.
Features:
Psychiatric Hospitalizations
Median Income by city district:
This database contains a compilation of all the New York City government-owned facilities. I wanted to see how many services the city offers to help with mental health treatment and also with employment or money management. I placed these locations on top of the choropleth graphs of the number of psychiatric hospitalization and median income. As such you can see which locations with high or low psychiatric incidences have certain mental health treatment options and also which locations with lower economic status have more workforce resources.
This entire data science project used python 3.6 with some additional modules installed. For data cleaning and engineering, I used the pandas module, using functions like read_csv() to open the csv file as a dataframe. Then I would clean and impute values to form a modified dataframe to use for my graphs. I used seaborn to make the summary statistics count plots and matplotlib to edit certain features like the xticks and yticks. I used folium to make my choropleth graphs with location tracking. Lastly, I used modules from sklearn to make my logistic regression and heat maps for my predictive plots.
This count plot shows the number of New Yorkers in 4 ordinal subgroups. Fewer people aged 18 to 24 responded to the survey but this group had the highest ratio of people with low mental health. All the other subgroups had an even distribution of low mental health and high mental health. There seem to be more occurrences of Normal mental health as the age increase.
Men had a lower ratio of low mental health than women. For women, it seems that the distribution is evenly split between normal mental health and low mental health.
Most people who were employed had normal mental health. However, those who were unemployed or not in the labor force had much greater incidences of low mental health. More unemployed people rated have low mental health. While those who are not in the labor force have an even split, half of them have a low mental state. These people include those who wish not to work and also those who cannot work at all.
New Yorkers who had an income less than 100% of the poverty level, which is $12,752 for a single, were more likely to have low mental health than not. As the income levels increase, you can see a weak trend that the occurrences of normal mental health increases.
The neighborhood poverty level distinguishes how many of the neighborhood population live below the poverty level. There does not seem to be a significant correlation between the neighborhood poverty level and mental health.
More people who missed rent had lower mental health, whereas more people who did not miss rent had normal mental health.
More people who cannot afford medication are likely to have low mental health.
There does seem to be a relationship between lower economic status and lower mental state. Those who are not able to afford their medication are more likely to have low mental health. The same is true for missing rent in the last 12 months and also by poverty level. This observation makes sense because missing rent, not being able to afford medication, and living below the poverty level is stressful and can be detrimental to mental health. In addition, a significant number of people who had low mental health were either unemployed or not in the labor force. These people who are facing mental distress or financial problems could benefit from certain programs. In the next part of the project, I will use a choropleth graph to map out locations by their number of psychiatric hospitalizations and the NYC mental health facilities. In addition, I will map out the locations of the NYC employment centers by areas with median income levels.
There are already many Mental Health Clinics that provide free or inexpensive services to help treat mental health. There are many clustered around areas of high psychiatric hospitalization and quite a few options to choose from. The total number of NYC government-owned facilities is 1,013. I had not known that the city provided so many resources for mental health. It is clear that NYC takes mental health seriously and is continuing to improve its infrastructure.
Many places with low median income have a few employment centers nearby. However, there are quite a few areas that have no employment centers at all. Yet, there are plenty of options to choose from as there are 145 of these centers in NYC. These employment centers can help New Yorkers find a new job or manage their money better all for free.
A few areas with very low median personal income are actually correlated with a greater number of psychiatric hospitalizations. Also, many high-income-earning areas are correlated with low psychiatric hospitalizations. While the community health survey took voluntary data from people, who were much less likely to face depression, New York open data reported tangible data on the number of psychiatric hospitalizations by borough. The lower-earning areas had greater mental breakdowns and hospitalizations, which does seem to link economic status with the occurrence of depression. As the data taken from the facility dataset shows, there are many cheap or inexpensive programs available to help people stuggling with their mental and financial health.
This logistic regression plot shows that one is slightly more likely to have low mental health with lower economic status and its probability decreases as you increase the economic status. You may be wondering why my logistic regression is a straight line. It does not appear as a sigmoid function because there are only 5 ordinal values instead of a spread-out distribution. Since the number of incidences of low mental health and normal mental health is close, the probability is not strong either way. This is also why the line is centered close to the middle, and also why the y value does not go beyond the threshold of 50%. There are just too many people who have normal mental health compared to low mental health. As such the logistic regression never predicts Low mental health (it never passes the threshold). We see this in the heatmap of the confusion matrix for this logistic regression.
The heatmap shows that the regression plot never predicts low mental health, always picking normal mental health. As a result, normal mental health is correctly predicted as normal mental health 54.24% of the time and incorrectly predicts low mental health as normal 45.76% of the time. The accuracy is then 54.24% and the precision is 0% since it never predicts low mental health. Since the data uses only categorical information with the low mental health and normal mental health having similar ratios of poverty groups, it is difficult to produce an accurate regression.
The summary statistics show that there is a slight correlation between economic status and mental health. The original idea was that poorer people would be less happy since they could not afford as much as more financially secure people. The count plots do show that people who have less money to afford medication and pay rent are likely to have lower mental health. They are more likely to report a low mood score or need mental health treatment. However, the data shows nearly the same amount of people reported to be depressed or not depressed regardless of their economic status. Because the difference observed is so slight, the logistic regression never predicts low mental health. Regardless, financial security plays a role in mental health and I analyzed the locations of nearby mental health service centers and employment centers by their psychiatric hospitalizations and median income level. I learned that there are many free and inexpensive resources for New Yorkers suffering from mental health problems and financial issues to use.
Depression is attributed to multiple causes, both genetic and environmental. Medication is one way to treat depression. While medicine only treats the symptoms of depression, other behavioral factors can also influence mental health. Such factors include diet and substance abuse. Further research can look into how these factors influence the precedence of mental health issues.