While we were off line ...

[AlanH]

To stave off the withdrawal symptoms during the forum upgrade, not only did I complete gotm31, but I finished off some work I'd been doing on extracting statistics from the GOTM results. You can now find a few graphs at http://gotm.civfanatics.net/graphs/display_graph.php or under the 'Statistics' link in the main GOTM site menu.

These are my first attempt at php graphics, so please be gentle. If you have any bright ideas for additional presentations, please let me know.

 

[scoutsout]

Those are pretty cool!

This would probably be a bit of a toughie... but what about breaking down the victory types by the year of victory? Showing the date range for each victory type... does that make sense? If you wanted to do it all on one graph you might be able to do it with lines instead of bars...

 

[AlanH]

Those are pretty cool!

This would probably be a bit of a toughie... but what about breaking down the victory types by the year of victory? Showing the date range for each victory type... does that make sense? If you wanted to do it all on one graph you might be able to do it with lines instead of bars...

I'll have a look at it. For each game or aggregate across all games? I know, I know ... Both

 

[mad-bax]

Yeah.... relative to "best" date from Jason would be good.

 

[Dianthus]

AlanH, thanks for the graphs. I've been going a bit graph mad recently. I made another one!

Dianthus' GOTM performance​

I was wondering if you could maybe add something like this? Bars rather than lines would be fine, and just one of the stats at a time. You would obviously also need a way to enter the player name as with the Global Rankings page.

What do you think?

 

[AlanH]

I was wondering if you could maybe add something like this? Bars rather than lines would be fine, and just one of the stats at a time. You would obviously also need a way to enter the player name as with the Global Rankings page.

What do you think?

Yes, not too difficult. There are various suggestions involving player ranking, and so far this is not a stored value in the database because the listings we have currently can do it easily with a sort. If we are going to want more per-player displays, it looks like I'll have to add it, so that will add a bit of work before I can graph it. My ToDo list is getting a little stretched right now, but I'll put this one on it.

A point of information:

I often see line graphs produced for the data sequences associated with games. You'll see that the graphs I have produces are all column format. This is not because I think they look prettier than lines. Bars or columns are the only valid output option for graphs where the X-axis is Game, because drawing a line implies there are, or could be, valid points on the graph along the lines. That's not true in this case, as game 28.5 never existed, and even if you speculate what might have happened if we had a game 28.5, there's no reason to believe the result would lie anywhere near the line between the 28 and 29 results. The result a player achieves in a game is affected by many variables, and the start date is only one parameter.

So this graph will use bars or columns. Here endeth the math lesson for today

 

[Dianthus]

Yes, not too difficult. There are various suggestions involving player ranking, and so far this is not a stored value in the database because the listings we have currently can do it easily with a sort. If we are going to want more per-player displays, it looks like I'll have to add it, so that will add a bit of work before I can graph it. My ToDo list is getting a little stretched right now, but I'll put this one on it.

You shouldn't need a ranking column for the percentage of the highest score for each month. I'm a bit of a dab hand with SQL if you want some help. It would make a nice change to write some useful SQL rather than the esoteric test SQL I mess around with at work!

 

[AlanH]

No, but you also want to graph your position as a percentage of lowest. I realise it's possible to extract that using SQL as well, but I just assume it would be more efficient to calculate rank once when we have the game results published and complete.