#UrbanDataCamp: Using Data to Help Low-Income Youth Find Jobs

Overview

Questions

Can we show that job opportunities for less-privileged youths leads to good outcomes later in life? (College enrollment, etc.)

Notes

General

"Alumni"
"Terminated" - didn't finish

Factors associated with a positive outcome:

college enrollment
college persistence
alumni

Methods

Load data

library(readr)
library(dplyr)

dat = read_csv('input/UA_UI_DataCamp_with_Codebook_4_29.csv')

## Warning: 2402 problems parsing
## 'input/UA_UI_DataCamp_with_Codebook_4_29.csv'. See problems(...) for more
## details.

# preview data
kable(dat[1:5, 1:6])

DataCamp ID	Academic Year	Location	Intern Status	Intern Work Site	Work Site ZIP Code
UADC1096	2014 - 2015	DC - Year Round High School Program	Active	Clark Construction Group, LLC Accounts Payable 2014-2015	20814
UADC1281	2014 - 2015	NCR	Active	Patent and Trade Office NCR 2014-2015	22314
UADC1129	2014 - 2015	Baltimore - YAIP	Alumni	ECSM BAL YAIP 2014-2015	21213
UADC1254	2014 - 2015	DC - Year Round High School Program	Active	Office of Personnel Management - Retirement Services 2014-2015	20240
UADC1007	2014 - 2015	NCR	Active	Alexandria City Attorney's Office NCR 2014-2015	22314

kable(dat[1:5, 7:12])

Organization	Industry	Total Hours Worked	DOB	Mailing ZIP/Postal Code	Median Household Income
Clark Construction Group, LLC	Real Estate	251	1996-02-21	20019	34832
Patent and Trade Office	Government	162	1997-02-15	22304	76061
Episcopal Community Services of Maryland	Non-profit	55	1993-11-15	21215	34471
Office of Personnel Management	Government	129	1997-10-18	20020	34685
Alexandria City Attorney's Office	Government	103	1997-06-11	22312	76845

kable(dat[1:5, 13:19])

Labor Force Participation	Unemployment Rate	SSI	Cash Assistance	SNAP	All Families Below Poverty	Educational Attainment - BA or Higher
0.595	0.136	0.115	0.103	0.324	0.264	0.134
0.788	0.045	0.022	0.017	0.048	0.046	0.586
0.555	0.102	0.123	0.064	0.268	0.223	0.171
0.561	0.115	0.113	0.124	0.348	0.317	0.160
0.757	0.037	0.013	0.016	0.056	0.091	0.435

kable(dat[1:5, 20:26])

Female-Headed Households	Race	Gender	High School	Title I School?	FRL	Graduation Rate of HS
0.188	Black	F	Washington Mathematics Science Technology Public Charter High School (WMST)	1	1.000	0.910
0.049	Hispanic/Latino	F	TC Williams	0	0.630	0.843
0.118	Black	F		NA	NA	NA
0.203	Black	M	Ballou Senior High School	1	0.998	0.500
0.044	Black	F	TC Williams	0	0.630	0.843

kable(dat[1:5, 26:33])

Graduation Rate of HS	College Enrollment Rate of HS	HS ZIP Code	1st Year College Enrollment	1st Year College Type (2/4)	1st Year College State
0.910	NA	20002	NA	NA	NA
0.843	NA	22302	NA	NA	NA
NA	NA	NA	NA	NA	NA
0.500	NA	20032	NA	NA	NA
0.843	NA	22302	NA	NA	NA

Variables

Note: Formatting for metadata is a little bit messed up...

metadata = read_csv('input/UA_UI_DataCamp_with_Codebook_4_29_metadata.csv')
kable(metadata)

Column Heading	Description
DataCamp ID	Unique intern identifier
Academic Year	Program year of participation
Location	Urban Alliance program/region
Intern Status	Active: Intern is currently in program
Alumni: Intern successfully completed pr	ogram.
Summer Program Alumni: Intern successful	ly completed the program and will likely return in the fall (Rising HS seniors)
Terminated: Dropped out of the program d	ue to performance reasons or self-removal
Intern Work Site	Tag for where interns worked, comprised of organization name, location, and academic year
Work Site ZIP Code	ZIP Code of Intern Work Site
Organization	Work Site Organization Name
Industry	Specific industry organization falls into
Total Hours Worked	Sum of monthly hours interns worked during the program; compiled from ADP payroll reports
DOB	Intern date of birth
Mailing ZIP/Postal Code	Intern's home ZIP Code
Median Household Income	ZIP Code-based; Median household income of residents within intern's home ZIP code; ACS 2013 5-year estimates
Labor Force Participation	ZIP Code-based; Labor force participation rate of residents within intern's home ZIP code; ACS 2013 5-year estimates
Unemployment Rate	ZIP Code-based; Unemployment rate of residents within intern's home ZIP code; ACS 2013 5-year estimates
SSI	ZIP Code-based; Percent of residents receiving SSI within intern's home ZIP code; ACS 2013 5-year estimates
Cash Assistance	ZIP Code-based; Percent of residents receiving cash assistance within intern's home ZIP code; ACS 2013 5-year estimates
SNAP	ZIP Code-based; Percent of residents receiving SNAP benefits within intern's home ZIP code; ACS 2013 5-year estimates
All Families Below Poverty	ZIP Code-based; Percent of residents living below poverty within intern's home ZIP code; ACS 2013 5-year estimates
Educational Attainment - BA or Higher	ZIP Code-based; Percent of residents with a Bachelor's degree or higher within intern's home ZIP code; ACS 2013 5-year estimates
Female-Headed Households	ZIP Code-based; Percent of residents living in female-headed households with children under 18 years of age within intern's home ZIP code; ACS 2013 5-year estimates
Race	Intern race (self-identified, voluntary)
Gender	Intern gender (self-identified, voluntary)
High School	High School intern attended
Title I School?	School-based
1: High School receives Title I funding
0: High School does not receive Title I	funding
blank: Title I data not found
FRL	School-based; Percent of students in high school qualified for free or reduced lunch
Graduation Rate of HS	School-based; Percent of high school students graduating within four years
College Enrollment Rate of HS	School-based; Percent of high school students who enroll in college following graudation
HS ZIP Code	ZIP Code of High School
NSC Data Available?	1: Alum has been included in a National Student Clearinghouse data request to determine college enrollment
0: Alum has not been included in a Natio	nal Student Clearinghouse data request to determine college enrollment
1st Year College Enrollment	Enrolled: Alum was enrolled in college at time of National Student Clearinghouse data pull
Unenrolled: Alum was not enrolled at tim	e of National Student Clearinghouse data pull
1st Year College Type (2/4)	2: Two-year college
4: Four-year college
No record found: Alum did not have a Nat	ional Student Clearinghouse record
1st Year College State	State in which college is located
Student Quote	Open-ended response from intern post-program survey

# convert dates to factors so they aren't treated as numeric
dat[['Work Site ZIP Code']]      = factor(dat[['Work Site ZIP Code']])
dat[['Mailing ZIP/Postal Code']] = factor(dat[['Mailing ZIP/Postal Code']])
dat[['HS ZIP Code']]             = factor(dat[['HS ZIP Code']])

table(dat[['Intern Status']])

## 
##                                 Active 
##                                    355 
##                                 Alumni 
##                                    744 
##                  Summer Program Alumni 
##                                     42 
##                     Terminated Student 
##                                     29 
## Terminated Student - Performance Based 
##                                    109 
##      Terminated Student - Self Removed 
##                                    117

Rename variables

For easier typing...

colnames(dat) = c('datacamp_id', 'academic_year', 'location', 'intern_status',
                  'intern_work_site', 'work_site_zip', 'organization', 'industry',
                  'total_hours_worked', 'dob', 'mailing_zip',
                  'median_household_income', 'labor_force_participation',
                  'unemployment_rate', 'ssi', 'cash_assistance', 'snap',
                  'all_families_below_poverty', 'educational_attainment_ba',
                  'female_headed_households', 'race', 'gender', 'high_school',
                  'title_i_school', 'frl', 'graduation_rate_of_hs',
                  'college_enrollment_of_hs', 'hs_zip_code', 'nsc_data_available',
                  'first_year_college_enrollment', 'first_year_college_type',
                  'first_year_college_state', 'student_quote')

# save with alternative name
write_csv(dat, 'input/UA_UI_DataCamp_with_Codebook_4_29_clean.csv')

Exploratory data analysis

Variable relationships

library(gplots)
library(bpca)
library(RColorBrewer)

# create short version of data with a subset of the fields and not including
# the current (2014-2015) interns who haven't yet finished; remove rows with
# duplicated ids
duplicate_ids = names(table(dat$datacamp_id)[table(dat$datacamp_id) > 1])
dat_short = dat %>%
    filter(academic_year != '2014 - 2015' & 
           industry != "" & 
           !datacamp_id %in% duplicate_ids) %>%
    select(datacamp_id, intern_status, industry, total_hours_worked,
           median_household_income, labor_force_participation,
           unemployment_rate, ssi, cash_assistance, snap,
           all_families_below_poverty, educational_attainment_ba,
           female_headed_households, graduation_rate_of_hs) %>%
    mutate(intern_status=ifelse(intern_status %in% 
                                c("Alumni", "Summer Program Alumni"),
                                "Alumni", "Terminated"))
    
dat_short_complete = dat_short[complete.cases(dat_short),]

ids = dat_short_complete$datacamp_id

dat_short_complete = dat_short_complete %>% select(-datacamp_id)

# Heatmap
heatmap.2(cor(as.matrix(dat_short_complete %>% select(-intern_status, -industry))),
          margins=c(26, 26), trace='none')

# group by location
dat$location_colors = brewer.pal(length(unique(dat$location)), "Set1")[
    as.numeric(as.factor((as.character(dat$location))))]
location_colors_complete = dat$location_colors[dat$datacamp_id %in% ids]

# biplot
plot(bpca(dat_short_complete %>% select(-intern_status, -industry),
          var.color=location_colors_complete, scale=TRUE))

Students

# rescale data
rescale = function(x) {
    (x - min(x)) / max(x)
}

dat_scaled = dat_short_complete

dat_scaled$median_household_income  = rescale(dat_scaled$median_household_income)
dat_scaled$total_hours_worked       = rescale(dat_scaled$total_hours_worked)
dat_scaled$ssi                      = rescale(dat_scaled$ssi)
dat_scaled$unemployment_rate        = rescale(dat_scaled$unemployment_rate)
dat_scaled$female_headed_households = rescale(dat_scaled$female_headed_households)
dat_scaled$cash_assistance          = rescale(dat_scaled$cash_assistance)
dat_scaled$snap                     = rescale(dat_scaled$snap)
dat_scaled$all_families_below_poverty = rescale(dat_scaled$all_families_below_poverty)

# numeric fields only
#mat = dat_short_complete %>% select(-intern_status)
#mat_scaled = scale(dat_short_complete)
heatmap.2(as.matrix(dat_scaled %>% select(-intern_status, -industry)),
          margins=c(26, 26), trace='none',
          RowSideColors=location_colors_complete)

Median household income vs. Hours worked

library(ggplot2)
ggplot(dat_short_complete, aes(total_hours_worked, median_household_income,
                               group=intern_status), color=intern_status) +
    geom_point(aes(color=intern_status))

Intern status vs. Industry

dat_short_complete %>% group_by(industry) %>% ggplot(aes(intern_status)) +
    theme(axis.text.x = element_text(angle = 90, hjust = 1)) +
    geom_bar() +
    facet_wrap(~industry)

#ggplot(dat_short_complete, aes(intern_status, total_hours_worked,
#                               group=industry), color=industry) +
#    geom_point(aes(color=industry))

Session info

date()

## [1] "Sat May  2 14:45:16 2015"

sessionInfo()

## R version 3.1.3 (2015-03-09)
## Platform: x86_64-unknown-linux-gnu (64-bit)
## Running under: Arch Linux
## 
## locale:
##  [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
##  [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
##  [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
##  [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
##  [9] LC_ADDRESS=C               LC_TELEPHONE=C            
## [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       
## 
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## other attached packages:
##  [1] ggplot2_1.0.1        RColorBrewer_1.1-2   bpca_1.2-2          
##  [4] rgl_0.95.1201        scatterplot3d_0.3-35 gplots_2.16.0       
##  [7] dplyr_0.4.1          readr_0.1.0          knitr_1.9.17        
## [10] rmarkdown_0.5.1      knitrBootstrap_1.0.0 setwidth_1.0-3      
## [13] colorout_1.0-3       vimcom_1.2-3        
## 
## loaded via a namespace (and not attached):
##  [1] assertthat_0.1     bitops_1.0-6       caTools_1.17.1    
##  [4] colorspace_1.2-6   DBI_0.3.1          digest_0.6.8      
##  [7] evaluate_0.6       formatR_1.1        gdata_2.13.3      
## [10] grid_3.1.3         gtable_0.1.2       gtools_3.4.2      
## [13] highr_0.4.1        htmltools_0.2.6    KernSmooth_2.23-14
## [16] labeling_0.3       lazyeval_0.1.10    magrittr_1.5      
## [19] markdown_0.7.4     MASS_7.3-40        munsell_0.4.2     
## [22] parallel_3.1.3     plyr_1.8.1         proto_0.3-10      
## [25] Rcpp_0.11.5        reshape2_1.4.1     scales_0.2.4      
## [28] stringr_0.6.2      tools_3.1.3        yaml_2.1.13

khughitt/urban-data-camp_youth-jobs

Overview

Questions

Notes

General

Factors associated with a positive outcome:

Methods

Load data

Variables

Rename variables

Exploratory data analysis

Variable relationships

Students

Median household income vs. Hours worked

Intern status vs. Industry

Session info