I am trying to visualize the data that I have recorded with a benchmark. The problem that i have is that the value changes are rather big and a single data set consists of 10.000 values. I tried to visualize it with python but the resulting line graph doesn't look good. Do you have some advice or ideas how i could visualize the data better?
No one can see 10K points on a single plot, therefore
- give the user the opportunity to display desired range
How is it possible? For example,
- user DataRangeSlider from bokeh widgets
- if your largest and smallest values are intrinsic to your system and you want to display them as they are, use logarithmic scale
- if your largest and and small values are outliers, use robust moving values, for example medians
- if your largest and and small values are outliers and you want to avoid averaging, just restrict the
y-axis to the values of interest
In my opinion I think it'll depend on why you want to visualize it / what you're trying to analyze. If it's appropriate, you might try something like a moving average to smooth out the sharp value changes.
I got this visualization from this blog post.
I found a another style of plot that might be interesting in this case: a boxplot.