--- tags: DSAI, python --- # Predict-Future-Sales Kaggle - 詳細報告：[report link](https://docs.google.com/presentation/d/1nMD0G22-CNBULoP0P4Gtn9MAQsO_bTo0vI2hzNKGz1k/edit?usp=sharing) - data [training](https://drive.google.com/file/d/1vOt8wdkSE-ulW5QVv6P2KDA83ARNWgrU/view?usp=sharing) [testing](https://drive.google.com/file/d/1ujWDRJztMpUUIK-m5dqlzNkmX_UXJ9E8/view?usp=sharing) ## 執行方式 使用 pipenv 作為虛擬環境 - 啟動虛擬環境 ```shell $ pipenv --python 3.8 ``` - 虛擬環境建立、指定版本為 python 3.8 ```shell $ pipenv shell $ (Predict-Future-Sales-Kaggle) chialiang86@chialiang86-System-Product-Name:~/Desktop/2021_DSAI/Predict-Future-Sales-Kaggle$ ``` - 套件安裝：附有 Pipfile, Pipfile.lock 檔案，直接安裝 ```shell $ pipenv install ``` :::info [packages] - numpy - pandas - seaborn - matplotlib - scikit-learn - lightgbm [requires] - python_version = "3.8" ::: ### 執行順序 提供兩個 .py 檔 - preprocessing.py : 做feature engineering，產生 training.pkl, testing.pkl 檔 ```shell $ python preprocessing.py ``` 執行完後會分別產生 training/testing 的 .pickle 檔及 .csv 檔 -> .pickle 為給 training.py 用 - training.py : 讀取 training.pkl, testing.pkl 進行訓練以及分析 ```shell $ python training.py ``` 執行後會產生 submission.csv