Prograss Report Zhe Chen 12-18-07

Finished creating Quality Control module, currently it can only track team killers and suiciders, as that's all the variables we can get.

Finished presentation slides for presentation.

Joe's Progress Report

This week I successfully ported the Learning component into the Matlab format.

Zhe Chen Progress Report

This is the progress report for last week.

-Finished porting Teambuilder algorithm from C++ to matlab. Input/Output interface is adjusted to fit with the output of linear regression and the input of the team creation functions.

Joe's Weekly Report

This week I met with Zhe to discuss porting our existing code to Matlab, and communicated with Rajesh vis-a-vis the procurement of mathematical modules in Matlab.

Mo - 9 December

Calvin - We'll need the documentation supporting the "TrueSkill" emulation, however it is it may operate.

Nathan - 12/8

Reporting a couple days late, but finished coding the "true skill" emulation in matlab. Although we claim it to be "true skill", it is but a guess at what it does, as the actual calculations are not available.

Ian's Progress Report, 12/4

- Finished Quality Control Module documentation

- Ran overnight bot test to produce large amounts of data

- Currently working on experimental explanation and procedure document

Joe's Progress Report 12/2/07

This week I coded the learning component of our project, integrated it with the parser, and met with Zhe to make sure that the matching algorithm will properly integrate with the learning component.

I noticed a bug in the parser which prevents it from working properly. Over the next week I plan to make sure that this bug is fixed, and to test the learning component with actual data.

Nathan - 12/2

Finalized linear regression by throwing away all the code and doing it in Matlab. Compared results against commercial software and they matched; a success in less than an hour's time.

Progress Report Zhe Chen 11-30-07

Had some problems with google account over the break, finally fixed it.
-Wrote C++ implementations of Matrix Multiplication and Matrix Transpose, along with a few useful Matrix related functions.
-Tested the algorithms with R generated matrixes
-Wrote Normalization algorithm for normalizing data before sending it to the Linear Regression Algorithm

Joe's Progress Report 11/26/07

This past week I created a new learning algorithm document which incorporated criticisms regarding the last one. I also updated the Makefile so that it will automatically compile the code required for the parser in addition to the code required for the matching algorithm. Over the coming week I plan to code the learning algorithm and integrate it with the matching algorithm and the parser.

Massive Update - Mo

Ok, A massive update was enacted earlier this week. The Wiki will be fully updated when we return on Tuesday, as I will need to make sure all the documentation is properly accessible on the Wiki, and Bin and I are going back and archiving a lot of information that was lost in the crash.

A large amount of documentation has been sent out to all of you, including an Excel sheet which constitutes our Gantt Chart.

Please read over the documentation and submit edits in the format I requested in the e-mail. Please DO NOT use MSFT Word's show edits feature as it is extremely difficult to use when multiple people edit a document.

DotProject is fully running now. DotProject is a collaborative software tool used mainly by LARGE-scale software project production, etc. Nevertheless I think it is a useful tool and hope it proves thus.

I have linked it off the Wiki: 160.39.129.104/dotproject

Your Login is your UNI, and the passwords are the same as the Wiki.

Professors and Lance, your login is "faculty" with the passwords also the same.

The software tracks dependencies, and notes progress and the interdependence of projects with each other.

Already some comments have been made on the Gantt Chart, and I will update it as necessary on Monday (I can't access my machine right now).

Testing is about to commence, and I will discuss with Mitchell a human playtest on 30November (Friday).

In addition I have been monitoring communication concerning the Linear Regression programming, and Bin and Zhe were able to test the matrix multiplication element.

Nathan Reports that his element is complete and requires a holistics check.

Bin and Calvin I will hold both of you to this task, as everyone else is very busy at the moment, and you're both intimately familiar with the regression and what it should do.

Calvin - I would also like a situation report concerning TrueSkill, given your recent revelation - that project is of extremely high priority.

Joe reports LA development is underway.

Rajesh, Ian and myself are a bit behind on the QC Module discussion, and will be meeting Monday or Tuesday latest for hopefully the last discussion - Zhe will then be tasked with coding it, as was discussed.

I hope you all have a good weekend, and as usual I sleep only every 2 days... so call me at ANY time with questions, concerns, etc.

Solo en muerte hace el fin de deber.

Nathan 11/19

Matrix class is completed. Linear Regression algorithm part1 is at testing phase to verify correctness.

Bin's Progress Report- 11/19

1- Set up a meeting on Wednesday (11/14/07) to prepare for the meeting on Friday. Redistributed assignments based on severity to get certain objectives completed. EX- added coding of Linear Regression as a top priority to get our system functional. Discussed briefly on the AI system.

2- Stayed the entire 13 hour long meeting on Friday (11/16/07) starting at 2pm till Saturday(11/17/07) 3am.

Crone Progress Report, 11/17

- Finished Variable Documentation
- Developed "Derived" Variables that might be more useful in matching than raw data, i.e. Reload Rate = (Reloads per round/Total number of shots in that round) rather than just the raw reload value.

Joe's Weekly Report 11/17/07

This week I rewrote the computation portion of the parser using Java. I worked with Rajesh and Calvin to make sure that the data exported by the parser is compatible with their requirements. I also wrote and presented a document for a proposed learning algorithm based on creating a neural network.

Over the next week, I plan to revise my parser code so that it will work well across platforms, and to create better documentation.


I also plan to revise my AI proposal based on the following criticisms:

Rajesh & Calvin: It's too impractical.

Jonathan: It's too generic.

Zhe Chen - 11/16/07

-Discussed with Calvin and Nathan on implementing Linear Regression, Covariance Matrix, and Data Normalization.
-Updated Teambuilder code to take dynamic player team sizes.

Nathan - Plugin 11/16

Blogspot wouldn't let me join the blog earlier during the week, which is the reason for the delay.

From 11/10, on the Wiki:

Successfully made the plugin dynamic in regards to the number of players, though there is still a hard cap at 5v5.

Successfully solved bugs related to memory crashes in attempting to count how often they have a particular gun out. Useful from a tendency point of view and skill point of view, as players run more quickly with a pistol,scout, or knife out.

Team in output should now work, geospatial errors in regards to player distance when hit/hitting should now be fixed.

11/16

Increased hard cap to 10v10.

Added handling of flashbangs and smokegrenades.

Solved an issue with the dedicated server and bots. They were joining twice according to the plugin, and that needed to be handled.

Output wasn't outputting properly (at all) for unknown reasons; had to modify the procedure to output the data after every round

Fixed problems related to uninitialized variables.

Sat with Calvin to discuss programming the linear regression to be run on data. Necessary functions have been covered, Zhe and I will be programming these functions by Sunday, then combining them to complete the algorithm by next week.

Calvin's Progress Report

I finished the report about linear regression design. Inside the report contains details about theory and facts about the design for finding the "good" secondary factors.

Bin's progress report

1- wrote graphical document explaining how our algorithm works

2-organized all meetings; entire group meetings and smaller group meetings.

3-some people in the group don't think on the same wavelength of what the project is suppose to look like or don't understand how specific aspects of the project works; so I set up meetings with a few team members to untangle inter-team confusion . example- made sure each member knew the difference between primary factors and secondary factors; clarified the AI module is different from linear regression analysis.

4- spent 30 straight hours with almost everyone in the group to make sure everyone was operating without conflicts.

5-took notes and minutes during meetings.

Progress Report

This past week I edited the Perl scripts for interacting with Nathan's plugin to be compatible with the revised data format.

This coming week, I plan to create up-to-date documentation for the scripts I've written, and create documentation for a proposed learning system. I also plan to come up with possible ways of using the other group to verify/refute the hypotheses we have created.

Ian's Progress Report, 11/11

1. Variable Document

I have continued to work on the variable explanation document, which will be ready for Friday's meeting.

2. Web Stuff

Installed first phpCollab and then dotproject for coordination purposes on the group site. Built a package of files as a backup for all of the group's work so far and to allow all group members to conduct testing.

3. QC Module

Met with Rajesh and Mo today to hammer out details of functionality of QC module, began to develop documentation on it. Asked Nathan to add 2 additional variables to the plugin, which he says is possible.

4. Additional documentation

Met with Jon to edit/revise presentation document.

Rajesh Progress Report

Accomplishments:

• I have done the "impossible" [or at least, what others have argued is impossible]: Developing the automation script that linked everything together.
• Worked with Zhe, Joe, Ian, and Nathan to connect and standardize everything
• Spent a full 30 hours working [and spent many hours trying to get people to finish their parts]

On the Docket

• Cleaning up the code base
• Preparing framework for the future steps
• Developing documentation

Progress Report: Zhe Chen 11-09-07

• The algorithm is now fully integrated with Joe's code on grabbing data from the database.
  
• Weights are now used to calculate the primary and secondary values, currently set to a set of default values.
• Algorithm can now take any number of players to insert into the player pool and grab data from.
• Outputs either verbose team makeup and data, or a teamlist that can be read by gsd.perl for automated team matching.
• Helped Rajesh integrate the entire automation process.
• Researched how to name/ID CS:Source Bots.
  

GSD Blog

Hello, This blog will be used to keep a steaming line of communication with everyone involved in the project. Everyone is encouraged to use it as a notification system, but it's not particularly effective as a chat medium, so I suggest using it for announcements, questions, etc.

I will be updating it with relevant information on a weekly, likely more often, basis, and would like everyone to subscribe to it, so that you're notified when changes are made.

It will be linked off of our Wiki for easy access as well.

PHP Collab isn't done yet, but I will notify when I've set it up.