I would like to use balnet in an application of the following type. I have a "sample" of units with features Xi, over i = 1, ..., n. I have a "population" of units with features Xj, over j = 1, ..., N. I would like to re-weight the "sample" so that its moments match those of the "population".
At first glance, this just looks like an application of balnet in ATT mode. However, there's a catch: N is very large, and also I can't actually query the individual Xj. All I have from the "population" is the output of a SQL query that gives me MU = 1/N \sum_{j = 1}^N Xj. (In contrast, I actually have the data from the "sample".)
It would be very helpful to have an alternate way to call into balnet for transport, that takes in the sample {Xi : i = 1, ..., n} as well as the target population mean MU.
I would like to use balnet in an application of the following type. I have a "sample" of units with features Xi, over i = 1, ..., n. I have a "population" of units with features Xj, over j = 1, ..., N. I would like to re-weight the "sample" so that its moments match those of the "population".
At first glance, this just looks like an application of balnet in ATT mode. However, there's a catch: N is very large, and also I can't actually query the individual Xj. All I have from the "population" is the output of a SQL query that gives me MU = 1/N \sum_{j = 1}^N Xj. (In contrast, I actually have the data from the "sample".)
It would be very helpful to have an alternate way to call into balnet for transport, that takes in the sample {Xi : i = 1, ..., n} as well as the target population mean MU.