physiology | R package to make physiology calculations

In fact, the first version, or a slight modification of it, is likely what you want here:

You should be aggregating only by the course name:

You need a join between the aggreated result and the original values based on the aggregated results

First, always reference column by name in aes and not by [ or $ of the same data frame. Second, you need to clean up your columns, ELEMENT_LABEL and plot_order, before plotting.

• Run droplevels as ELEMENT_LABEL currently has 54 levels for only 28 observations! Likely, your plotting data frame derives from a filtered version of original. After doing so, you reduce to 8 levels across 28 observations.
• Re-assign values of plot_order column by the unique values, not simply an integer sequence across length of rows. Specifically, plot_order must order the 8 unique values to correspond to each unique occurrence of ELEMENT_LABEL. You can achieve this with inline aggregation like ave calling min to take lowest number for same ELEMENT_LABEL.

Once you clean up these two columns, plotting then should adjust:

You can access the stats included in the boxplots. Here's some illustrative data:

Following a dialog with Google, it turns out the error was caused by a broken image URL. I was using the direct pixabay URL. Changing that to download the image and serve it from my system. solved the problem.

It might be easier to parse your input using preg_match_all; this can give you arrays of depths and associated values. You can then iterate through those arrays, opening an

when the depth increases, and closing it when the depth decreases:

I'm guessing that when you say "mydata is very very huge", that means it is much bigger than the example data you've given in your question. Otherwise it is hard to imagine this taking minutes to run.

The algorithm can be improved by iterating through the list once, instead of just once per keyword. The trick is to test whether an item matches one or more keywords using a set intersection. In order to build separate result lists per keyword, we can store them in a dictionary. It's also only necessary to call literal_eval once per item, even if it matches multiple keywords.

Updated:

One solution would be to use join with dplyr: