Welcome to the 2nd Annual HelpMeViz Vizathon event! I am again happy to be working with the Bread for the World Institute on issues related to hunger in the developing world. This year, we’re also teaming up with the International Food Policy Research Institute, who is providing one of their newest datasets on hidden hunger.
We also have a second team of volunteers in San Francisco, led by Leigh Fonseca of Living Data. They will link up with us at 12:30ET/3:30PT from their site at the headquarters of Macys.com in San Francisco to share ideas.
Here’s the deal: the data for this event are available on this Github page, so grab it, download it, fork it, whatever. And get started! Drafts, images, and questions can be posted in the comments section below or on Twitter using the hashtag #hiddenhunger.
Images from the event, conversation summaries, and important notes will be posted throughout the day on this site and on Twitter.
Thank you for participating in this very important event and contributing your time, skills, and expertise to helping those in need.
![Spotify](https://policyviz.com/wp-content/uploads/2023/09/main-og-image.png")
Here is the link to the mapping tool from Harvest Choice: http://apps.harvestchoice.org/mappr/
And here’s the Harvest Choice table tool: http://harvestchoice.org/page/tablr
We think there may be some issues with the weight and height variables in the Ethiopia dataset on Github. Malawi seems okay.
Okay, we’ve resolved this:
Age is in months
Weight is in kilograms
Height is in cm
Great work from DC to checkout : http://keshif.me/demo/hiddenhunger
What do “rec” and “req” mean (Malawi data)? I’m thinking the recommended level would be higher than the required level, and so I would never expect to see rec as No and req as Yes, but it happens often. For example, first row for Vita_C. The codebook description of the fields suggest Yes means deficient: “consumes less than the rec/req amount of X”. Or make rec and req stand for different words…
https://raw.githubusercontent.com/jgaw/nutrition_indicators/master/correlations.csv
Malawi – BMIZ06, WHZ06, WAZ06, HAZ06 correlation coefficients with every other variable to help with building graphs.
https://github.com/jgaw/nutrition_indicators
Now updated to get p-values and correlations
https://github.com/jgaw/nutrition_indicators/blob/master/correlations_and_Pvals.csv
Some basic data exploration and looking into gender discrepancies in micro-nutrient deficiencies :
https://github.com/khughitt/hidden-hunger-hackathon-2015
Here we look at the distribution of diets in Malawi, normalized to the median pulse consumption of households experiencing a calorie deficity
Here, we list the weighted percentages of the Malawi data set for the percentage of the population experiencing a given nutritional deficiency…
calcium 99.2
fol 85.2
iron 87.1
kcal 78.6
nia 53.7
prot 60.5
ribof 45.0
thia 35.4
vita_A 57.7
vita_B12 74.6
vita_B6 50.5
vita_C 41.3
vita_E 80.7
zinc 88.3
William, I read it as the other way around. 99% not calcium deficient. However, see my question above indicating my confusion about those variables.
Some rather horrible python code for exploring the Malawi data set:
import sys, os
import numpy as np
import pandas as pd
import pylab
if __name__ == “__main__”:
filedir=r”/Users/williamratcliff/nutrition_indicators/Malawi”
nutrition=os.path.join(filedir,”mwi_nutrition.csv”)
anthropometry=os.path.join(filedir,”mwi_anthropometry.csv”)
df=pd.read_csv(nutrition)
df_an=pd.read_csv(anthropometry)
df_join=pd.merge(df,df_an,on=’hhid’)
dependent=’tot_iron_cons’
independent=’ddi’
independent=’wealth’
independent=’def_kcal_ae_hh’
i=1
dlist=[‘tot_iron_cons’,
‘tot_cu_cons’,
]
dlist=[‘sh_roots_kcal’,
‘sh_veg_kcal’,
‘sh_fruits_kcal’,
‘sh_meat_kcal’,
‘sh_eggs_kcal’,
‘sh_fish_kcal’,
‘sh_pulses_kcal’,
‘sh_milk_kcal’,
‘sh_oil_kcal’,
‘sh_sugar_kcal’,
‘sh_misc_kcal’]
if 0:
for dependent in dlist:
pylab.subplot(4,4,i)
pylab.plot(df_join[independent], df_join[dependent],’ro’)
pylab.xlabel(independent)
pylab.ylabel(dependent)
i=i+1
if 0:
df_join[dlist].plot(kind=’bar’, stacked=True);
if 1:
norm=df_join[df_join[‘def_kcal_ae_hh’]==1][dlist[6]].median()
for i in range(len(dlist)):
pylab.subplot(2,1,1)
pylab.bar(i, df_join[df_join[‘def_kcal_ae_hh’]==1][dlist[i]].median()/norm,color=’blue’)
pylab.text(7.5,0.8, ‘Calorically deficient’)
for i in range(len(dlist)):
pylab.subplot(2,1,2)
pylab.bar(i, df_join[df_join[‘def_kcal_ae_hh’]==0][dlist[i]].median()/norm,color=’red’ ) 5,0.8, ‘Calorically sufficient’)
pylab.xticks(range(len(dlist)), dlist)
pylab.title(‘Diet distribution in Malawi’)
pylab.show()
This rough Tableau visualization draft may point the way towards a path for further analysis. It takes the Top and Bottom villages and ranks them by different vitamin deficiencies, controlled by a parameter (It filters out villages with 50 or less households). It then displays the average share of what foods are consumed in each of these villages.
https://public.tableau.com/views/HiddenHungerVizathon-TopandBottomMallawiVillagesBasedonVitaminDefficiencies-Whataretheyeating/TopandBottomVillagesBasedonVitaminDef?:embed=y&:showTabs=y&:display_count=yes
Nice meeting you all today! I have learnt a lot from all of you.
Here is a summary of what I did today on the Malawi data about household wealth, education level and deficient micro-nutrients:
http://angelcymak.github.io/Vizathon_hiddenHunger/
Nice meeting you all today, I learned so much. Thanks for showing up and sharing ideas and knowledge.
Here is the link to quick tableau viz.
Summary of further exploration:
1) EPA area where there are most stunted Children – Kandeu
2) More than half of populations are stunted. There is no significant difference between male and female. Although, there seem to be more significant stunted male in household in highest wealth quantile
3) The more children in household (2-5) children, more likely to be stunted
4) Speaking English (more stunted?)
5) Polygamous household shows more stunted children
https://public.tableau.com/profile/prapakarn#!/vizhome/Malawi-Vizathon2015/Story1
https://public.tableau.com/profile/prapakarn#!/vizhome/Malawi-Vizathon2015/Story1
Findings:
— Male children are more likely to be wasted than female children
–This is true across all age groups for children
–Infants are most likely to be wasted.. Is this because they are more limited in the types of food they can consume vs children who can chew and thus have more options?
We see also differences in wasting based on the head of household and whether the household is in a rural area or small town
Infographic from #HiddenHunger Vizathon from Becca Rhodes (https://twitter.com/BeccaStreets)
http://helpmeviz.com/wp-content/uploads/2015/06/HiddenHunger.png