Targeting and Fairness Data#
Example file: synthetic_data/targeting.csv
unique_id |
weight |
outcome |
col1 |
… |
col35 |
|---|---|---|---|---|---|
jb9f3j |
12.6 |
1.5 |
5.2 |
… |
category3 |
ad8b3b |
1.6 |
2.7 |
6.1 |
… |
category6 |
oewf9ras |
3.2 |
4.1 |
1.5 |
… |
|
bncz83 |
6.0 |
0.9 |
2.6 |
… |
category2 |
Fields
#
*unique_id: string
Unique ID for each survey observation
weight: float
Weight for each survey observation (if sample weights are used)
outcome1…outcome20: float
“Ground truth” poverty outcomes. Can have any column names.
proxy1…proxy20: float
Poverty proxies used for targeting. Can have any column names.
characteristic1…characteristic20: object
Characteristics for fairness analysis. Must be of categorical type.
Warning
All proxy measures should go in the same direction!
Note
Columns without a preceding asterisk ‘*’ are optional