MLGlobalHealth/Realtime-SAE-Zimbabwe

# Realtime-SAE-Zimbabwe This is a repository for the work "Real-time small area estimation of Food security in Zimbabwe: integrating mobile-phone and face-to-face surveys using joint multilevel regression and poststratification." (arXiv). ## Structure of the repository The repo should have a `data` folder and a sub-folder `raw` with the following datasets: * Mobile-phone data: In our analysis, WFP's mVAM survey for Zimbabwe conducted within Real-time Monitoring system * Face-to-face data: In our data analysis, this is The Rural Zimbabwe Vulnerability Assessment Committee (ZimVAC) survey. See Food and Nutrition Council Zimbabwe for the related reports. We used Household-level data (2023 survey for modelling, 2024 survey for validation).The aggregated data (2021 and 2022 survey) are used only for reference. Both data should have the same structure, e.g. | household_id | adm1_name| adm2_name | obs_month | fcs | covariate_1 | covariate_2| $\dots$ | | -----------: | --------:| --------: | -----: | ----------: | ----------:| ------: |------: | | 1 | Matabeleland South | Beitbridge | 2023-05-01 | 40 | A | B | In addition, we will also need * Data from the 2015 Zimbabwe DHS survey which can be can be requested here. * Zimbabwe MICS6 SPSS Datasets. This can be requested and downloaded from here. * Nigeria's shapefile: `Zimbabwe_adm2_uscb_2022.shp` can be downloaded from here. This is modified so that the `adm2_name` matches those from mVAM surveys, and saved to `processed` as `shapefile.rds` * Zimbabwe 2022 census: `n_hh_by_district_and_ward.xlsx` should contain information extracted from section 5 of the report. The data should be organized in the following tree structure: ``` ├── data | ├── raw | │ ├── mVAM | | | ├── WFP_mvam_weights_pop.xlsx | | | └── mVAM.csv | │ ├── ZimVAC | | | ├── zimvac2024_survey.csv. | | | ├── zimvac2023_survey.csv | | | ├── zimvac2022_agg.csv | | | └── zimvac2021_agg.csv | │ ├── zimbabwe-adm1 …