jimver / prettypercentiles

Python utility for pretty percentile plots

prettypercentiles

Python utility for pretty percentile plots. Ideal for plotting latency graphs from csv files.

Usage

Install Python 3 (Tested so far with 3.9)

(Optional but recommended to prevent cluttering your global pip) create a virtual environment in this directory:

python -m venv .venv

Activate the virtual environment:

.venv\Scripts\activate

Install dependencies from requirements.txt:

pip install -r requirements.txt

Run plotter:

python prettypercentiles.py -c configs.example_config

Configuration

Configuration files are located in configs/

The configuration file is just a Python file with variables. If you want to make your own copy the example_config.py to a new file in the same folder.

There are comments for each setting, the most important ones are these:

tuple of (csv filename, column name, preprocessing function)

These define where and which column to get and an optional preprocessing function to run it through

hgrm_map

A dict which maps the above columns to labels in your plot. The key is the label in the plot and the value is the file name to the hdrhistogram output.

label_map

A dict which maps the above columns to labels in your plot. The key is the label in the plot and the value is the tuple

combined_columns

A dict that combine several columns into one using a given function. The key is the label in the plot and the value is a list of 2 or more tuples

Note that the hgrm labels cannot be combined as they contain the percentiles and not the latencies.

num_intervals

Number that determines how far into the nines you want to plot. For example: 3 means you will plot the following intervals:

[0-90, 90-99, 99-99.9]

So the higher the number, the closer to 100% you will plot.

font_scale

Float that increases/decreases font size. 1.0 is normal size. For example presentations usually have higher font size compared to papers.

Output

Running example config:

python prettypercentiles.py -c configs.example_config

the above command results in the following image in images/example_plot.png:

Dark mode config (white text and grid lines):

python prettypercentiles.py -c configs.example_config_dark

output image in images/example_plot_dark.png: