Lab #1

[jasminacosta]

Hi @JasonSills!

For part 4 on the lab I am a little bit confused on the questions and how to proceed answering them.

• Q4b: What is the counterfactual? Provide both its formula and estimation.

• [ ] To find the counterfactual, do we need to look from a certain point of time? Or right after the intervention begins?

• Q4c: What would the counterfactual be after 150 days?

• [ ] For this counterfactual is it asking about day 150 , or 150 days after the intervention?

• Q4d: Are the two counterfactuals the same? Why?

• [ ] Is this comparing if both counterfactuals are the same when looking at the same time to see if it is the same number of passengers?

Thank you!

[JasonSills]

@jasminacosta

For Q4b you need the equation for the counterfactual and you can use the time at the intervention.

For Q4c the counterfactual is after the intervention. It is what would have occurred had their not been the intervention.

For Q4d compare the counterfactual at the intervention and 150 days after the intervention.

[jasminacosta]

@JasonSills

Understood, thank you!

[jasminacosta]

@JasonSills

Another question: When we are estimating the counterfactual for different times within the time series model, are we estimating the counterfactual based on where the point is plotted? Or do we need to write an equation to make that estimation for us?

[JasonSills]

@jasminacosta

The estimation at that point in time will work.

[jasminacosta]

@JasonSills I apologize just to clarify, we are eyeballing the estimation of the counterfactual based on the point that is plotted on the graph? We do not need to create an equation to try and solve that?

[JasonSills]

@jasminacosta -

Not eyeballing it. You need to use code and your regression equation to state the predicted value at a given time. For example, let's say at time X, what is the counterfactual:

data_X_C <- as.data.frame(cbind( Time = X, Treatment = 0, TimeSince = 0)) # Data if the intervention did not occur, so Treatment and TimeSince are both 0

passengers_X_C <- predict(regTS, data_X_C)
passengers_X_C

We walked through this in the live session that was recorded and sent through announcements, so they might help.

[jasminacosta]

@JasonSills

Oh, I see!

I just did not print my assignment to be able to see my value.

Thank you for the clarification.