This module visualises Age of Empires II DE statistics of replays segmented into multiple skill levels:
- PRO (ELO > 2200)
- HIGH (ELO > 1800)
- MID (ELO > 1000)
- LOW (ELO ≤ 1000)
It also generates CSV files for each replay file calculated for each player. This can be deactivated by commenting line 337 in analysis.py.
The replays are added to ez-aoe-details/replays. The current replays have been taken from aoe2.net and all feature games played as 1v1 on Arabia. Graphs generated with the given replay files can be found in ez-aoe-details/graphs.
It aims to understand how player behaviour varies based on skill level, for me a more thourough explanation, see Problem Description and Summary.
Players are able to understand their own actions in game by analaysing their replay files. In particular, the tool enables them to understand their weak points.
git clone https://github.com/juliastic/ez-aoe-details.gitpip3 install pandaspip3 install matplotlibpip3 install git+https://github.com/happyleavesaoc/aoc-mgz.git@fix-fast
python3 ez-aoe-details/analysis.py
Data is passed as dictionary through the different classes. All player relevant data is stored in a player.Player instance. The players are held in their respective analysis.Analysis classes which are again part of analysis.MultipleAnalyses. All project relevant constants are stored in units.py, i.e., technologies, units and buildings. The data is visualised in visualisation.AoEGraphs.
The replay files are parsed seperately in analysis.Analysis and the actions are added to their players. This is done by checking the ID of the action and handling it accordingly, For example, in case of a unit queue, the unit ID is parsed. In player.Player, the data is matched with its name and stored accordingly. Every two minutes, the data is calculated and stored in a dictionary entry:
{0: {'units': {Economy: 0, Military: 0}, 'buildings: {Economy: 0, Military: 0}, 'action_move_coordinates': {'x': 0, 'y': 0, count: 0}, 'action_move_coordinates': {'x': 0, 'y': 0, count: 0}}}.
After all analyses have been concluded, the data from all players from the same skill levels are averaged and a dictionary with relevant calculations is created in analysis.MultipleAnalyses:
{'low': {120: {'units': {Economy: 4.8, Military: 0.05}, 'buildings': {Economy: 0.1, Military: 0.05, Wall: 0.0}, 'action_move_coordinates': {'x': 17.178742296106446, 'y': 24.073642806206504, 'count': 564}, 'action_unit_coordinates': {'x': 0.0, 'y': 0.0, 'count': 0}}}}.
The key is a specific timestamp, the subkeys store the average amount of TOTAL created units/buildings up to the timestamp and the action location between the last timestamp and the current one timestamp. Count is required for the coordinate calculations since all coordinates are added and later divided by their number of occurence (key count).
In order to calculate the average research time of technologies and effective actions per minute (i.e., moving units, attacking), a similar approach is utlised. The average research time for all players is stored in seconds. In the end, the total resarch time is divided by the number of players who have researched the technology:
{'eapm': {'pro': 45.921025117599676, 'high': 44.40773592782665, 'middle': 29.72721939895853, 'low': 16.29198645698805}, 'technologies': {'pro': {Economy: {Loom: 367.25, Feudal Age: 463.6, Double Bit Axe: 639.8421052631579}}}.
All relevant averaging calculations happen in analysis.MultipleAnalyses.compute_average_results_for_type(). The data points are then visualised in the graphs.
- Average eAPM
- Average villagers queued over the course of a game with average age up times
- Average villagers queued over the course of a game with key economy research
- Average military units queued over the course of a game with key military research
- Average military buildings built over the course of a game
- Average pallisade walls built over the course of a game
- Average occurence of attacking actions in game
- Average distance of moving actions from starting position
- Average distance of attacking actions (i.e., formation, attacking or patrol) from starting position
- Military count for all timestamps
- Villager count for all timestamps
- Technology research time
- Queued units
The calculations are done for each player.
The generated CSV files are added to output/timestamp/REPLAY_FILE.
The parser can be used with any replay files from the game Age of Empires II DE.
To customise the analysed replays, the main method in ez-aoe-details/analysis.py can be adapted. Notably, constants SkillLevel.CUSTOM_A and SkillLevel.CUSTOM_B have been added to facilicate this action. They can be used when referencing paths containing other replays:
analyses = MultipleAnalyses({SkillLevel.CUSTOM_A: file_path + '/custom_a', SkillLevel.CUSTOM_B: file_path + '/custom_b'})
To amend the visualisations, file visualisation.py needs to be modified. Modification of displayed technologies can be easily changed by amending the technologies below comment # set relevant technologies. For an overview of currently implemented technologies see also units.Technology.
- Abstract visualsations to enable easier reuse, i.e., enable users to decide which technologies to visualise
- Add missing IDs: Not all unit and technology IDs have been added as they were not relevant for the selected replay files
- Extend CSV implementation
- Add pip package support