Getting Better at Data Analysis 1

Working on my Data Analysis project yesterday, I realize my current EDA and Data Analysis project is so inefficient.

First, I am spending too much time trying to make a particular graph, using a technology(ggplot2) that I am not familiar with. This distracts me from my analysis process. Also in terms of making graphs, I have to get familiar ggplot2, and learn how to make informative graphs. Googling “how to make XX using ggplot2” is very tedious and inefficient, besides the trial-and-error of making better graphs.

Being good takes hours and hours of deliberate practice. I think there are a few important metric for being better at this:

1. How much I am learning from the best?
   Roger Peng’s video is pretty helpful. Maybe I should check datatau more. Also Andrew Ng’s course on Structuring Machine Learning Project definitely worth a second visit. Reading EfficientR is super good. It taught me many helpful technical tips, such as benchmarking. I tested library Tidyverse’s mutate function on transforming a column, it gives me 10x(3s) performance, comparing to using a base R function sapply;I also tried compiling the mapping function but didn’t very little in that case.
2. How much time I spent streching my limits? Need to force myself from making poor decisions, such as overspending my time on making beautiful plots before finishing my analysis;also need to force myself to learn ggplot2 comprehensively, reduce amount of time I have to google to make a plot.