Download DHS data

Country selected: Kenya

Import the households dataset for your selected country and create a data frame with a variable that describes each of the following: household ID, unit, weights, location, size, gender, age, education, and wealth. Pivot the persons columns within your households data to a long format in order to produce a similarly specified dataset that describes all persons residing within all households.

Code used:

``` households <- read_dta(“KEHR72DT/KEHR72FL.DTA”)

hhid <- households$hhid #check length(unique(hhid)) unit <- households$hv004 weights <- households$hv005 / 1000000 location <- as_factor(households$shcounty) size <- households$hv009 sex <- households[ ,312:334] age <- households[ ,335:357] edu <- households[ ,358:380] wealth <- households$hv270

hhs <- cbind.data.frame(hhid, unit, weights, location, size, sex, age, edu, wealth)

gender <- hhs %>% pivot_longer(cols = starts_with(“hv104”), names_to = “pid”, values_to = “gender”, values_drop_na = TRUE)

age <- hhs %>% pivot_longer(cols = starts_with(“hv105”), names_to = “pid”, values_to = “age”, values_drop_na = TRUE)

edu <- hhs %>% pivot_longer(cols = starts_with(“hv106”), names_to = “pid”, values_to = “edu”, values_drop_na = TRUE)

gender <- select(gender, -starts_with(“hv”)) age <- select(age, -starts_with(“hv”)) edu <- select(edu, -starts_with(“hv”))

gender$id <- paste(gender$hhid, substr(gender$pid, 7,8), sep = ‘’) age$id <- paste(age$hhid, substr(age$pid, 7,8), sep = ‘’) edu$id <- paste(edu$hhid, substr(edu$pid, 7,8), sep = ‘’) pns <- merge(gender, age, by = ‘id’) pns <- merge(pns, edu, by = ‘id’) ```

Using this data frame describing all persons standardize, normalize and percentize your variables and visualize each post transformed dataset as a heatmap that illustrates the heterogeneity of the combination of patterns.

Raw heatmap:

Scaled heatmap:

Normalized heatmap:

Percentized heatmap: