Team Fight Tactics Strategy Application

http://tftchamp.duckdns.org:3000/

Datasets

publish @ https://www.kaggle.com/datasets/teckmengwong/team-fight-tactics-matches

Requirements

To develop and use this code, you will need:

a Riot Games Developer account at https://developer.riotgames.com/
a Riot Games API key
a mongodb services

Create and store in tftchamp/.env (user environment in Windows will overwrite)

RIOT_API_KEY for Riot API key. ATLAS_URI for mongodb connect uri. DB_NAME default db.

ATLAS_URI=mongodb+srv://<username>:<password>@sandbox.lqlql.mongodb.net/?retryWrites=true&w=majority
DB_NAME=tftchamp
RIOT_API_KEY="RGAPI-blah-blah-blah"

DEV env

Python >= 3.10
Packages included in requirements.txt file
(Anaconda for easy installation)

Python virtual env setup

For local setup, I recommend to use Miniconda, a minimal version of the popular Anaconda distribution that contains only the package manager conda and Python. Follow the installation instructions on the Miniconda Homepage.

After installation of Anaconda/Miniconda, run the following command(s) from the project directory:

Install dependencies

Conda virtual environment:

conda create --name myenv python=3.10
conda activate myenv
conda install --file requirements.txt -c conda-forge

As Conda has limited package support for python 3.10 activate your virtual environment then install the dependencies using

pip install -r requirements.txt

Usage for Jupyter Notebook

Activate and install the correct python3 virtual environment before proceeding.

jupyter notebook

Datasets

About this dataset

Team Fight Tactics highest ELO challengers games scrape by https://github.com/furyhawk/tftchamp. Using https://developer.riotgames.com/ API.

8 players FFA in one game.

Target Label: placement 1 is best. Lower is better. Top 4 placement is a Win. Alternative prediction is to group Top 4 placement as Binary Win, bottom 4 as Binary Lost.

Only team traits and augments/items chosen included in datasets. Stats like game_length, players_eliminated are excluded. This is to prevent the model from learning obvious predictor.

Challenges:

Different baseline between games(match_id). Different games has different baseline at different ELO. This may affect the true value of Global placement.
Categorical values.

## Example
{'match_id': 'EUW1_5944909932', #Match id.
 'placement': 2, # target_label.

# max 3 augments player can choose at stage 2-2, 3-2, 4-2.
 'augment0': 'TFT6_Augment_TrueTwos',
 'augment1': 'TFT6_Augment_CyberneticImplants2',
 'augment2': 'TFT6_Augment_SecondWind2',

# traits
 'Set7_Assassin': 1,
 'Set7_Bruiser': 1,
 'Set7_Cavalier': 2,
 'Set7_Dragonmancer': 0,
 'Set7_Guild': 3,
 'Set7_Legend': 0,
 'Set7_Mirage': 0,
 'Set7_Ragewing': 1,
 'Set7_Swiftshot': 1,
 'Set7_Tempest': 1,
 'Set7_Warrior': 1,

# units/items
 'TFT7_Sejuani': 2,
 'TFT7_Qiyana': 2,
 'TFT7_Shen': 2,
 'TFT7_Shen_item0': 'TFT_Item_ChainVest',
 'TFT7_Shen_item1': 'TFT_Item_Shroud',
 'TFT7_Twitch': 2,
 'TFT7_Twitch_item0': 'TFT_Item_MadredsBloodrazor',
 'TFT7_Hecarim': 2,
 'TFT7_Talon': 2,
 'TFT7_Talon_item0': 'TFT_Item_RedBuff',
 'TFT7_Talon_item1': 'TFT_Item_InfinityEdge',
 'TFT7_Talon_item2': 'TFT7_Item_CavalierEmblemItem',
 'TFT7_Xayah': 2,
 'TFT7_Xayah_item0': 'TFT_Item_GuardianAngel',
 'TFT7_Xayah_item1': 'TFT_Item_InfinityEdge',
 'TFT7_Xayah_item2': 'TFT_Item_GuinsoosRageblade',
 'TFT7_Ornn': 2,
 'TFT7_Ornn_item0': 'TFT_Item_GiantsBelt',
 'TFT7_Ornn_item1': 'TFT_Item_Shroud',
 'TFT7_Yasuo': 1,
 'TFT7_Yasuo_item0': 'TFT_Item_RapidFireCannon',
 'TFT7_Yasuo_item1': 'TFT_Item_MadredsBloodrazor',
 'TFT7_Yasuo_item2': 'TFT_Item_TitansResolve'
}

**MetadataDto**
Name 	Data Type 	Description
data_version 	string 	Match data version.
match_id 	string 	Match id.
participants 	List[string] 	A list of participant PUUIDs.

**InfoDto**
Name 	Data Type 	Description
game_datetime 	long 	Unix timestamp.
game_length 	float 	Game length in seconds.
game_variation 	string 	Game variation key. Game variations documented in TFT static data.
game_version 	string 	Game client version.
participants 	List[ParticipantDto] 	
queue_id 	int 	Please refer to the League of Legends documentation.
tft_set_number 	int 	Teamfight Tactics set number.

**ParticipantDto**
Name 	Data Type 	Description
companion 	CompanionDto 	Participant's companion.
gold_left 	int 	Gold left after participant was eliminated.
last_round 	int 	The round the participant was eliminated in. Note: If the player was eliminated in stage 2-1 their last_round would be 5.
level 	int 	Participant Little Legend level. Note: This is not the number of active units.
placement 	int 	Participant placement upon elimination.
players_eliminated 	int 	Number of players the participant eliminated.
puuid 	string 	
time_eliminated 	float 	The number of seconds before the participant was eliminated.
total_damage_to_players 	int 	Damage the participant dealt to other players.
traits 	List[TraitDto] 	A complete list of traits for the participant's active units.
units 	List[UnitDto] 	A list of active units for the participant.

**TraitDto**
Name 	Data Type 	Description
name 	string 	Trait name.
num_units 	int 	Number of units with this trait.
style 	int 	Current style for this trait. (0 = No style, 1 = Bronze, 2 = Silver, 3 = Gold, 4 = Chromatic)
tier_current 	int 	Current active tier for the trait.
tier_total 	int 	Total tiers for the trait.

**UnitDto**
Name 	Data Type 	Description
items 	List[int] 	A list of the unit's items. Please refer to the Teamfight Tactics documentation for item ids.
character_id 	string 	This field was introduced in patch 9.22 with data_version 2.
chosen 	string 	If a unit is chosen as part of the Fates set mechanic, the chosen trait will be indicated by this field. Otherwise this field is excluded from the response.
name 	string 	Unit name. This field is often left blank.
rarity 	int 	Unit rarity. This doesn't equate to the unit cost.
tier 	int 	Unit tier.

Database services ./tftchamp/backend/mongodb/

Set your Atlas URI connection string as a parameter in .env. Replace and with your credentials. Create and store in tftchamp/backend/mongodb/.env

ATLAS_URI=mongodb+srv://<username>:<password>@sandbox.lqlql.mongodb.net/?retryWrites=true&w=majority
DB_NAME=tftchamp

Localhost mongodb

Or local hosting using docker-compose file @ backend/mongodb/docker-compose.yml

etc. create .env ./tftchamp/backend/mongodb/

MONGO_ROOT_USER=admin
MONGO_ROOT_PASSWORD=1234
MONGODB_DATABASE=tftchamp
MONGODB_LOCAL_PORT=27017
MONGODB_DOCKER_PORT=27017
MONGOEXPRESS_LOGIN=user
MONGOEXPRESS_PASSWORD=1234
MONGOEXPRESS_LOCAL_PORT=8888
MONGOEXPRESS_DOCKER_PORT=8081

to start the service

cd backend/mongodb
docker-compose up -d

Application service WIP @ ./tftchamp/backend/app/

We'll use the python-dotenv package to load environment variables ATLAS_URI and DB_NAME from the .env file. Then, we'll use the pymongo package to connect to the Atlas cluster when the application starts. We'll add another event handler to close the connection when the application stops.

Create and store in ./tftchamp/backend/app/.env .env

ATLAS_URI=mongodb+srv://<username>:<password>@sandbox.lqlql.mongodb.net/?retryWrites=true&w=majority
DB_NAME=tftchamp
RIOT_API_KEY="RGAPI-blah-blah-blah"
PATCH=2023-05-16
LATEST_RELEASE=13.10.509.8402

uvicorn main:app --host 0.0.0.0 --port 8000
# python3 -m uvicorn main:app --reload

When the application starts, navigate to http://localhost:8000/docs and try out the match endpoints.

Machine Learning Pipeline ./tftchamp/backend/pipeline/

Create and store in ./tftchamp/backend/pipeline/.env .env

ATLAS_URI=mongodb+srv://<username>:<password>@sandbox.lqlql.mongodb.net/?retryWrites=true&w=majority
DB_NAME=tftchamp
RIOT_API_KEY="RGAPI-blah-blah-blah"
PATCH=2023-05-16
LATEST_RELEASE=13.10.509.8402

scrape
data_loading
team_composition ~~4. data_analysis.ipynb~~
optimizer.py
Prediction service TODO

cd backend/pipeline
python3 scrape_db.py -c configs/challengers.json --no-load_new
python3 scrape_db.py -c configs/challenger_oce.json --no-load_new
python3 scrape_db.py -c configs/grandmasters.json --no-load_new
python3 data_loading_db.py -c configs/challengers_loading.json
python3 data_loading_db.py -c configs/grandmasters.json
python3 team_composition_db.py -c configs/challengers_loading.json
python3 team_composition_db.py -c configs/grandmasters.json
python3 optimizer.py -c configs/config_tft.json

windows scripts

scripts/run_new.bat
scripts/run_refresh.bat
scripts/run_preprocess.bat

linux/macOS scripts

# or for grandmaster and above league only
bash scripts/run_pipeline.sh [-h] [-nri] [-mgc] [-a] -- Team Fight Tactic pipeline helper

where:
    -h  show this help text
    -n  Scraping for new summoners (default=0)
    -r  Scraping for new matches (default=1)
    -i  Transforming and infering matches (default=0)
    -m  run above for master league (default=0)
    -g  run above for grandmaster league (default=0)
    -c  run above for challenger league (default=1)
    -a  run all above (default=0)

Config file format

Configure the config files under backend/pipeline/configs folder. Config files are in .json format. Example of such config is shown below:

{
    "name": "Classification",   // session name

    "model": {
        "type": "Model",    // model name
        "args": {           // args to be pass to class init
            "pipeline": ["scaler", "PLS", "pf", "SVC"],     // pipeline of methods
            "unions": {     // unions of methods included in pipeline
            }
        }
    },

    "tuned_parameters":[{   // hyperparameters to be tuned with search method
                        "SVC__kernel": ["rbf"],
                        "SVC__gamma": [1e-5, 1e-6, 1],
                        "SVC__C": [1, 100, 1000],
                        "PLS__n_components": [1,2,3]
                    }],

    "optimizer": "OptimizerClassification",    // name of optimizer

    "search_method":{
        "type": "GridSearchCV",    // method used to search through parameters
        "args": {                  // args to be pass to class init
            "refit": false,
            "n_jobs": -1,
            "verbose": 2,
            "error_score": 0
        }
    },

    "cross_validation": {
        "type": "RepeatedStratifiedKFold",     // type of cross-validation used
        "args": {                              // args to be pass to class init
            "n_splits": 5,
            "n_repeats": 10,
            "random_state": 1
        }
    },

    "data_loader": {
        "type": "Classification",      // name of dataloader class
        "args":{                       // args to be pass to class init
            "data_path": "../data/noshow.db",    // path to data
            "shuffle": true,    // if data shuffled before optimization
            "test_split": 0.2,  // use split method for model testing
            "stratify": true,   // if data stratified before optimization
            "random_state":1    // random state for repeated output
        }
    },

    "label_name": "no_show",       // specify the label column name
    "score": "max recall",     // mode(max min) and metrics used for scoring - https://scikit-learn.org/stable/modules/model_evaluation.html#scoring-parameter
    "train_model": true,    // if model is training
    "test_model": true,     // if model is tested after training
    "debug": false,         // debug model architecture
    "save_dir": "saved/",   // directory of saved reports, models and configs
    "model_dir": "saved/Classification/0115_140140"         // directory to load saved model. Leave empty "" to auto load current training model.
}

Scraping script ./tftchamp/backend/pipeline/scrape.py

For scrape->data_loading->team_composition challengers.json is used.

{
    "name": "challengers",              // config folder to creat under `save_dir`
    "load_new": true,                   // scrape, whether to fetch new summoners
    "servers": ["na1", "euw1", "kr"],   // regions to `scrape`, `load`, `process` ['euw1', 'na1', 'kr', 'oc1']
    "league":  "challengers",           // league to `scrape`, `load`, `process` 'challengers', 'grandmasters'
    "max_count": 30,                    // max matches per `scrape`
    "latest_release": "12.14.455.6722", // game version for cutoff '12.12.450.4196' '12.13.453.3037' Version 12.12.448.6653 12.11.446.9344 Version 12.13.453.3037
    "ranked_id": 1100,                  // `1090` normal game `1100` ranked game
    "patch": "2022-07-27",              // patches released date(2022, 7, 1) date(2022, 7, 16)
    "debug": false,                     // not used
    "save_dir": "saved/"                // `process stage`: output to
}

To run with custom arg --no-load_new:

cd backend/pipeline
python3 scrape_db.py -c configs/challengers.json --no-load_new

Front End REACT app ./tftchamp/frontend

DEV test

npm install # once only
npm start

Production LIVE

docker-compose up

Credits

Riot API discord https://discord.gg/riotgamesdevrel

https://www.mongodb.com/atlas/database

Mandu icons created by Freepik - Flaticon Tribal icons created by Freepik - Flaticon Electronic icons created by Freepik - Flaticon Brain icons created by Freepik - Flaticon Code icons created by Freepik - Flaticon Artificial intelligence icons created by Freepik - Flaticon Healthcare and medical icons created by Freepik - Flaticon Games icons created by Freepik - Flaticon Fire icons created by Freepik - Flaticon Unicorn icons created by Freepik - Flaticon

Recent rankings

2023-05-16 v12.19

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
.github/workflows		.github/workflows
assets		assets
backend		backend
frontend		frontend
scripts		scripts
tests		tests
.dockerignore		.dockerignore
.gitattributes		.gitattributes
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
SECURITY.md		SECURITY.md
create_virtualenv.sh		create_virtualenv.sh
docker-compose.yml		docker-compose.yml
main.yml		main.yml
tft.yml		tft.yml

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

Team Fight Tactics Strategy Application

Datasets

Requirements

Python virtual env setup

Install dependencies

Usage for Jupyter Notebook

Datasets

About this dataset

Challenges:

Database services ./tftchamp/backend/mongodb/

Localhost mongodb

Application service WIP @ ./tftchamp/backend/app/

Machine Learning Pipeline ./tftchamp/backend/pipeline/

Config file format

Scraping script ./tftchamp/backend/pipeline/scrape.py

Front End REACT app ./tftchamp/frontend

Credits

Recent rankings

top5_items

kmeans_comp_ranking13.10.509.8402_2023-05-16

dbscan_comp_ranking

augments_stage2-1_ranking

augments_stage3-2_ranking

augments_stage4-2_ranking

About

Releases

Packages

Languages

License

furyhawk/tftchamp

Folders and files

Latest commit

History

Repository files navigation

Team Fight Tactics Strategy Application

Datasets

Requirements

Python virtual env setup

Install dependencies

Usage for Jupyter Notebook

Datasets

About this dataset

Challenges:

Database services ./tftchamp/backend/mongodb/

Localhost mongodb

Application service WIP @ ./tftchamp/backend/app/

Machine Learning Pipeline ./tftchamp/backend/pipeline/

Config file format

Scraping script ./tftchamp/backend/pipeline/scrape.py

Front End REACT app ./tftchamp/frontend

Credits

Recent rankings

top5_items

kmeans_comp_ranking13.10.509.8402_2023-05-16

dbscan_comp_ranking

augments_stage2-1_ranking

augments_stage3-2_ranking

augments_stage4-2_ranking

About

Topics

Resources

License

Security policy

Stars

Watchers

Forks

Releases

Packages 0

Languages

Packages