search

Sayani07 / paper-gravitas

Exploring probability distributions for bivariate temporal granularities
0 stars 0 forks source link

fig 6 doesn't really work for the purpose #10

Open dicook opened 4 years ago

dicook commented 4 years ago

you can two granularities both with length 3, so its hard too see straight off without reading the caption that these are the same granularities re-mapped. Would be better to have one with different number of categories. Why did you just use a subset of dow and moy?

It's good that it's a small number, and so plots are large, though.

It might be good to have a pair showing a clash too, despite what we discussed in the last meeting. With a manageable subset like this its possible to expand with a small clash example.

I wouldn't mention the Victorian electricity data here. It's a basic example of how a harmony or a clash evolve, and we aren't learning anything about Vic electricity. It could confuse the reader about the later example in the application example.

When you rearrange to a different combination change the colour palette. This is another indication that its a different set of granularities shown. Using the same palette for two different plots indicates to the reader that they need to be comparing the colours across the displays, which is not what you want.

dicook commented 4 years ago

Oh, because this is in effect a small "simulated" example, you could modify the data so that there is more difference between the distributions in the different arrangements of granularities.

Sayani07 commented 4 years ago

Thanks for your suggestions. I had taken a subset of moy and dow so that there are fewer categories to compare. Now I have changed the granularities to be weekend/weekday and quarter-of-year so that we have fewer and different number of categories. Also added a basic example of clash with quarter-of-year and month-of-year.