Project 2
Project 2
Project Details
Project 2 was to analyze and create a few machine learning models against the online news popularity data. The analysis separated the data into subsets by the “data_channel_is” columns. So, for each of these columns we subsetted the data and ran the analysis against that data. Then, we did this five more times with the other “data_channel_is” columns. For example, one column is “data_channel_is_lifestyle” and it will be 1 with each of these that are about lifestyle. We did the analysis on all data where this is true and then continued to do the same analysis with the other columns.
The project site is here and the repo is here.
Difficult Parts
Perhaps the most difficult part for me was getting the automation
correct. It took a few attempts to first make sure I passed in the data
channels as a parameter correctly, but also to make sure it was using
the parameters to subset the data correctly. I kept trying to use a
string to subset the data like it was a column name similar to
news %>% filter("data_channel_is_lifestyle" == 1) but that kept
erroring out. Instead I realized that I can use the bracket notation and
pass the string in that was like in
news %>% filter(news["data_channel_is_lifestyle"] == 1). Once I got
that the automation was pretty much done.
Take-aways
The main take-away was that I realized how powerful R Markdown can be, especially when you can pass parameters into it. I’ll definitely need to keep this in mind if I need to create analysis reports again.