#datamanagementandvisualization

Starting with the importing of the necessary libraries, choosing a theme and making sure our text gets printed out nicely on the plots:

Afterwards, making a histogram on the left and a boxplot on the right, showing median, IQR, and outliers (one plot per variable):

For a finish, using matplotlib to create a scatterplot/regression, including a Pearson correlation.

The results are as follows:

The results from this analysis do not find any significant correlation between employment rate and alcohol consumption.

Everyone drinks!

This would serve as a continuation of my last post concerning the frequency distributions.

i continue my code by cleaning the columns and making sure I only leave in the values that are ≥ 0 for alcohol consumption and [0, 100] for employment rate.

Then I add two boolean flags: True if the cleaned value is not missing, False otherwise.

Then print everything:

Afterwards, I convert continuous variables into easy, human-readable groups (into quartiles) with the help of bins:

Afterwards, I define a helper that prints a clean frequency table for categorical data (or anything treated as categories plus a row for NaN because dropna=False):

Finally, just print everything:

The results are as follows:

This serves as a very nice continuation of everything we have done thus far.

As per my last post, I did a frequency distribution between employrate and alcconsumption using Python.

This code: loads the file, makes sure both variables are numeric, builds a binned frequency table (counts, %, cumulative %) and prints missing values.

I have also created a version full with comments explaining how it works:

The results we get is two clean frequency distribution tables:

This type of analysis is a small but crucial step we take before diving deeper into our analysis.

After looking through the codebook for the Gapminder study, I noticed myself particularly interested in the connection between alcohol consumption and employment rate. So I decided to use exactly those corresponding variables (alcconsumption, employrate) in my own codebook.

My interest stems from both alcohol dependent people dealing with their unfortunate lack of employment, as well as those dealing with their unfortunate poor working environments. There are a lot of studies that have already tackled the effect of unemployment on alcohol consumption. It has been found to have positive links in alcohol consumption rate (Popovici et al., 2013). De Sio et al. (2020) have also found that office work reduces the chances of a person developing an alcohol dependence.

What I want to take a look at is how alcohol consumption and employment rates differ across countries and how it correlates with low and high employment rate regions.

Popovici, I., & French, M. T. (2013). Does unemployment lead to greater alcohol consumption?. Industrial Relations: A Journal of Economy and Society, 52(2), 444-466.

1.- CODE

LIBNAME MILIBRO "/home/u59780920/sasuser.v94" /*access=readonly*/; DATA MISDATOS; set MILIBRO.MISDATOS; /*libraryNAME.dataNAME*/ /*Labeling the relevant variables*/ LABEL ID="Unique ID" S1="Sex" S3A="Living with her mother (2 means YES)" S3B="Living with her father (2 means YES)" S3A1="Living with her mother only" S3B1="Living with her father only" S7M="Mother's prof. situation" S8="Mother's prof. category" S81="New mother's prof. categorization" S34="Wished study level"; /*Set missing data (9 means EMPTY)*/ IF S1=9 then S1=.; IF S3A=9 then S3A=.; IF S3B=9 then S3B=.; IF S7M=9 OR S7M=6 then S7M=.; /* 6 means I DON'T KNOW*/ IF S8=99 then S8=.; IF S34=9 then S34=.; /*Creating secondary variables*/ IF S3A=. then S3A1=.;/*If the main variable is empty*/ IF S3A=2 AND S3B=1 then S3A1=2;/*If lives with her mother and not with her father*/ ELSE IF S3A=2 and S3B NE 1 then S3A1=1; /*If lives with her mother and with her father or there is no info about him*/ IF S3A=1 then S3A1=1;/*If does not live with her mother*/ IF S3B=. then S3B1=.;/*If the main variable is empty*/ IF S3B=2 AND S3A=1 then S3B1=2;/*If lives with her father and not with her mother*/ ELSE IF S3B=2 and S3A NE 1 then S3B1=1; /*If lives with her father and with her mother or there is no info about her*/ IF S3B=1 then S3B1=1;/*If does not live with her father*/ /*Grouping the variable "S8-Mother`s Professional Category" into less categories*/ IF S8=1 OR S8=2 then S81=5;/*Prof. category that require University degree*/ IF S8=3 OR S8=4 then S81=4;/*Prof. category that require High Vocational Training*/ IF S8=5 OR S8=6 then S81=2;/*Prof. category that require Medium Vocational Training*/ IF S8=7 OR S8=8 then S81=3;/*Prof. category that require Bachelor*/ IF S8=9 then S81=1;/*Prof. category that require Mandatory studies*/ IF S8=. then S81=.; /*Only girls are part of this study and 1 means girl*/ IF S1=1; PROC SORT; by ID; /*Sorted by the unique student ID*/ /*Frequency tables*/ PROC FREQ; tables S1 S3A S3B S3A1 S3B1 S7M S8 S81 S34; /*Data tables for checking*/ /*PROC PRINT; var S8 S81; PROC PRINT; var S7M S81;*/ RUN;

2.- RESULTS

STEP 1

First data management decision has been to code out missing data in all affected variables.

Later I have created secondary variables obtaining when the girls live with their mother only or with their father only, creating the variables S3A1 and S3B1.

Finally, as there are many options for the question “Mother professional situation” (variable S8), I have decided to make smaller groups, grouping those professional categories that imply similar level of studies, creating the new variable S81. Initially there were 10 different categories (including empty-99) and I have grouped them in 5 + the empty category (.). Trying to make it easier to compare, I have used the same logic than the Level of Wished Studies (variable S34) answered on the next question, i.e. if the mother’s professional category implies university studies then the classification for the new variable S81 is 5, and so on.

STEP 2

In the frequency distribution tables it is interesting to see that 98% of girls live with their mother (11.5% with her only) compared to the 87% that live with their father (1.3% with him only). (See below tables 1 and 2)

Regarding the new variable S81, around 60% of mothers do a job that only requires mandatory or basic studies (1-35.4% and 2-25.6%) and there is only a 10% of mother doing a job with university degree. (See Table 3)

1) your program

/*As I’m using my own data, I need to Initialize data from a .xlsx file. Only needed to be done once to create the library. Find the code below*/

LIBNAME MILIBRO "/home/u59780920/sasuser.v94";

FILENAME RUTA '/home/u59780920/sasuser.v94/EGD10.xlsx';

PROC IMPORT DATAFILE=RUTA

               DBMS=XLSX

               OUT=MILIBRO.MISDATOS;

               GETNAMES=YES;

RUN;

PROC CONTENTS DATA=MILIBRO.MISDATOS;

RUN;*/

 /*Now the code for the program itself, once the library has been created*/

LIBNAME MILIBRO "/home/u59780920/sasuser.v94";

DATA MISDATOS; set MILIBRO.MISDATOS;  /*libraryNAME.dataNAME*/

LABEL    ALUMNO="Unique ID"                      /*Labeling the relevant variables*/

                              S1="Gender"

                              S7M="Mother prof. situation"

                              S34="Wished study level";

IF S1=1;                                                             /*Only girls are part of this study --> 1 = girl*/

PROC SORT; by ALUMNO;                            /*Sorted by the unique student ID “Alumno”*/

PROC FREQ; tables ALUMNO S1 S7M S34; /*To display the distribution of each variable*/

RUN;

 2)  the output that displays three of your variables as frequency tables

 (see pictures) 

3)  a few sentences describing your frequency distributions in terms of the values the variables take, how often they take them, the presence of missing data, etc.

The variables frequency distribution shows on the first table (GENDER) that the filter is properly working: we want to study only the answers from girls (13830).

Then, the second variable frequency distribution shows that almost 60% of the mothers works outside the home (1), around 6% works from home (2), 6.8% are unemployed (3), a 0.5% are retired (4), a 22.8% are housewives (5), there are a 1.14% of girls that do not know what their mother do for living (6) and finally, there are 2.6% of missing data (9)

During selection of relevant data sets I found most interesting  AddHealth code boom that represent information about different aspects of life teenagers in 7-12 grade.

I would like to focused on  Personality and Family  information and study following questions:

- What is correlation between communication with parents and life satisfaction , personality and  patterns of decision-making

- How the origin of child affect of relationship with a parents as part of parents culture.

To answer these question I am taking data from 2 section

- Section 1: General Introductory

- Section 18: Personality and Family

Variable Name Description of Indicator Main Source Step 1:After reviewing the code book for gap minder i have decided to study the relationship between the employment and suicide rate, so i will be adding employrate, suicideper100TH into my personal code book.

Step 2:Unemployment among youth can cause demotivate which can lead to depression and eventually suicide. BY conducting this study i would like to study on the correlation between the two variables.

Step 3:

As a second topic i  would like to study the relationship between  polityscore and  employrate. so i am adding polityscore also to my personal codebook