Build SCM Dashboard

[AaronCooke2718]

https://media.discordapp.net/attachments/1034287087458996244/1226295972758294648/20240406_180909.jpg?ex=6624401f&is=6611cb1f&hm=dc18ca16e35f70ba2d2020c99226274eece4d43adf051d2a2414c313e3f1e156&=&format=webp&width=591&height=788

2 pages, organized into a data selection and model results page.

For data selection, choosing either raw variables, and/or factors generated by the DFM. The you choose which states you want to include, and have a preview table to review data.

Then you select which variable for the outcome (for us, pandemic intensity will probably be the goto) as well as which variables are predictors (a collection of factors and other econ variables as controls)

You choose which state is your treatment state (A state with high vaccination, long school closure, etc) and when the treatment period began (March 2020)

Then you run model

On the results page you can see a plot of the real state outcome variable and the synthetic variable.

DATA SELECTION

#SELECT H5AD FILE (DEFAULTS TO FILE USED FOR LAST DFM RUN)

#SELECT VARIABLES FOR PREVIEW TABLE
#DROP DOWNS TO SELECT SERIES FROM H5AD FILE
#CAN ADD OR REMOVE DROP DOWNS
#ALSO BUTTON TO ADD EVERY SERIES
#CAN ADD STATES OR LEAVE SOME OUT
#BUTTON TO GENERATE PREVIEW TABLE

#SELECT VARIABLES FOR MODEL
#SELECT Y (OUTCOME VARIABLE)
#SELECT TREATMENT STATE
#SELECT K PREDICTORS
#SELECT TREATMENT PERIOD
#BUTTON TO RUN MODEL

#RESULTS

#PLOT OF Y AND Y^
#VECTOR OF Wj

[jvivian]

@AaronCooke2718 — Beautiful thanks. Currently, we have one $t$ x $F$ matrix for every batch run $b$. Can you describe the structure of the data used as input to the SCM?

[jvivian]

@AaronCooke2718 — When you can take a look, based on what I'm seeing this is a non-issue as we need to specify an ID-var anyway so the SCM expects the data in one big matrix with an ID discriminator:

https://github.com/OscarEngelbrektson/SyntheticControlMethods/blob/master/SyntheticControlMethods/main.py#L283-L303

[AaronCooke2718]

@jvivian I agree, the ID (he calls it unit) variable will be State, correct?

[jvivian]

Ya, I think so! So the matrix will be time by factors with an added State var column, which I believe is already the output, I'll double-check.

[jvivian]

Closed via #82