Skip to content

Instantly share code, notes, and snippets.

View mathesong's full-sized avatar

Granville Matheson mathesong

32 followers · 29 following
View GitHub Profile
@mathesong
mathesong / Power_Question.md
Last active November 9, 2023 19:27
Power Question: multiple noisy measurements vs single noiseless outcome

Aim

I want to simulate to demonstrate how it's possible that the hierarchical model can even have better power than getting exactly the right values.

Note: The original question and results are no longer applicable due to a few fatal flaws with my code. Below follows the correct code. But the oracle is optimal now...

Libraries




  


  




    


          
    

      

        
        

          

              @mathesong
          

          

            
                mathesong
                / mgcv_predict1000
            
            

              Created
              February 19, 2019 11:24
            

              
                I cannot predict more than 1000 points using the mgcv package at a time, otherwise it starts repeating the values
              
          

        

      

        

  

    
    


        



  


  

        

          
          library(mgcv)
        

        

          
          library(tidyverse)
        

        

          
          

        

        

          
          # Data
        

        

          
          

        

        

          
          BPR <- 
        

        

          
            tibble::tribble(
        

        

          
            ~time,              ~bpr,
        

        

          
            50, 0.892042350231095,
        

        

          
            170, 0.888132434012036,
        

  





    


  




    


          
    

      

        
        

          

              @mathesong
          

          

            
                mathesong
                / reliability_gif.R
            
            

              Created
              November 15, 2018 17:15
            

              
                Code to make a gif demonstrating varying degrees of reliability, and the amount of measurement error relative to the true variation in the data. This comes from trying to explain concepts in my preprint here: https://www.biorxiv.org/content/early/2018/06/05/274894
              
          

        

      

        

  

    
    


        



  


  

        

          
          library(tidyverse)
        

        

          
          library(magick)
        

        

          
          

        

        

          
          # devtools::install_github("thomasp85/patchwork")
        

        

          
          library(patchwork)
        

        

          
          

        

        

          
          colours = c("#85d4e3",
        

        

          
                      "#e39f85")
        

        

          
          

        

        

          
          set.seed(40)
        

  





    


  




    


          
    

      

        
        

          

              @mathesong
          

          

            
                mathesong
                / reminderbot.R
            
            

              Last active
              November 9, 2023 19:16
            

              
                Making a Reminder Bot for Automating Meeting Organisation with R and Google Sheets
              
          

        

      

        

  

    
    


        



  


  

        

          
          # Please note that this script is not fully tested. This is a genericised 
        

        

          
          # version of two working versions, where I've added and removed bits and 
        

        

          
          # pieces along the way without testing them. Please let me know if there 
        

        

          
          # are errors or things missing, because this could easily have happened.
        

        

          
          

        

        

          
          

        

        

          
          library(tidyverse)
        

        

          
          library(googlesheets)
        

        

          
          library(lubridate)
        

        

          
          library(gmailr)