Hello all,
The last few days I have been putting together a pair of scripts for taking daily league snapshots and calculating various statisitics from them. I have a small homepage for the utility.
The basic set-up is in two parts:
1. The downloader: grabs each of the league web pages on a daily basis and saves them. Extracts the data from the html and saves it as a csv file for each league.
2. The analyser: takes the data from the csv and outputs statisitics.
The downloader is operational and I have been 'officialy' grabbing data from the 1st May (although I do have a few days before that as well).
I am currently in the process of writing the analysis functions. Once that is done, I shall wrap everything up in a usuable UI, stablise the code (bits are a tad flaky) and possibly generate some html/text reports instead of raw stats.
Does anyone have any ideas for stats/analysis that I can output? At present I have the following functions:
1. take a single data set and return win/draw/lose ratios for each player.
2. take two data sets and return the differences in each of the league values (rank, rating, played, win, draw, lose) for the intersection of players in both data sets.
3. take six data sets (the three leagues on two different dates) and return the promotions and demotions between each league, new players that have joined and old players that have dropped off.
I am considering another that calcualtes a simple rating points per game stat.
I think that things such as filtering come more under the remit of the UI and report design, but I intend to include them.
In the fullness of time I will release the code to one and all, but at the moment it is a bit messy and fudged in places, so I want to tidy it up first.
I will also offer to run any analysis on data that I have on request (if anyone is really that fussed), until I can either automate it from a proper web facing database or release a decent UI version that can grab back data from somewhere.
There isn't too much value in this project, I can't see many people beating down my doors for the output, but it keeps me amused and sharpens my coding a bit.
All thoughts welcome (except the ones that say it is apointless waste of time, I know that already).
bmd
The last few days I have been putting together a pair of scripts for taking daily league snapshots and calculating various statisitics from them. I have a small homepage for the utility.
The basic set-up is in two parts:
1. The downloader: grabs each of the league web pages on a daily basis and saves them. Extracts the data from the html and saves it as a csv file for each league.
2. The analyser: takes the data from the csv and outputs statisitics.
The downloader is operational and I have been 'officialy' grabbing data from the 1st May (although I do have a few days before that as well).
I am currently in the process of writing the analysis functions. Once that is done, I shall wrap everything up in a usuable UI, stablise the code (bits are a tad flaky) and possibly generate some html/text reports instead of raw stats.
Does anyone have any ideas for stats/analysis that I can output? At present I have the following functions:
1. take a single data set and return win/draw/lose ratios for each player.
2. take two data sets and return the differences in each of the league values (rank, rating, played, win, draw, lose) for the intersection of players in both data sets.
3. take six data sets (the three leagues on two different dates) and return the promotions and demotions between each league, new players that have joined and old players that have dropped off.
I am considering another that calcualtes a simple rating points per game stat.
I think that things such as filtering come more under the remit of the UI and report design, but I intend to include them.
In the fullness of time I will release the code to one and all, but at the moment it is a bit messy and fudged in places, so I want to tidy it up first.
I will also offer to run any analysis on data that I have on request (if anyone is really that fussed), until I can either automate it from a proper web facing database or release a decent UI version that can grab back data from somewhere.
There isn't too much value in this project, I can't see many people beating down my doors for the output, but it keeps me amused and sharpens my coding a bit.
All thoughts welcome (except the ones that say it is apointless waste of time, I know that already).
bmd
--
LSN Name: Big Mad Drongo (Elite)
Alliance Clan: Rock 'N' Roll Suicides (Bass Player)
LSN IRC: #lasersquadnemesis on irc.quakenet.org
LSN Name: Big Mad Drongo (Elite)
Alliance Clan: Rock 'N' Roll Suicides (Bass Player)
LSN IRC: #lasersquadnemesis on irc.quakenet.org
