# SML Final Project

Subject: Prediction for heat flux on dataset "playground-series-s3e15"

Author:

## Data Preparation

Get data from:
 - https://www.kaggle.com/datasets/saurabhshahane/predicting-heat-flux
 - https://www.kaggle.com/competitions/playground-series-s3e15/data

And place them under the `data` directory.

## Build Running Environment

We use docker to build the running environment. If you don't have docker on your windows yet, you may should install wsl2 + docker for windows first.

Once the docker backend is ready, place yourself in the base directory of this project, and run the following command to build the docker image:

```bash
docker build -t sml-final-project:flag .
```

To run the docker container with your local gpu available and local directory mounted, you can use the following command:

```bash
docker run -itd --gpus all -v $(pwd):/mnt/sml sml-final-project:flag bash
```

Now the docker container is running in the background. You can use the following command to check the container status:

```bash
docker ps
```

And to enter the container:

```bash
docker exec -it sml-final-project bash
```

You will see "root" as user name, which means the env is now ready. You may need to manually install some packages list below:
 - seaborn
 - ipykernel
 - jupyter
 - lightgbm
 - xgboost
 - catboost
 - sdf (maybe older version than 0.3.6 is required for numpy compatibility)

## Run the project's training and eval

## Run the project demo

Normally in vscode, you need to install "jupyter" extension to detect the python kernel. You can either select the python interpreter directly or install a new jupyter kernel with the following command:

```bash
python -m ipykernel install --user --name sml-env --display-name "Python (sml-env)"
```

how to use look up table

```bash
cd thirdparty/2006_Groeneveld_CriticalHeatFlux_LUT && python 2006LUT_to_SDF.py
```

```bash
python test/test_lut.py
```