R Shiny app for visualisation of the 1-step estimator for the Average Treatment Effect (ATE). Before using the app it is strongly recommended that you read the two excellent papers
- Visually Communicating and Teaching Intuition for Influence Functions by Fisher and Kennedy (https://arxiv.org/abs/1810.03260)
- Demystifying statistical learning based on efficient influence functions by Hines et. al. (https://arxiv.org/abs/1810.03260).
We hope that this app can serve as a complement to these papers and provide further intuition of how the influence function can be used, specifically in the context of the ATE and the 1-step estimator.
We hope that this app can provide further intuition of how the influence function can be used, specifically in the context of the ATE and the 1-step estimator.
Please visit the app on shinyapps.io to try it out online.
Clone the repo and open server.R or ui.R in RStudio and press the "Run App" button. You may have to install one or several dependencies before being able to run the app.