Time Series Train/Test Split
Time series train/test splits must preserve chronological order. The article uses a 75% training and 25% test split. Shuffling time series data with `train_test_split` causes data leakage. Data leakage can make evaluation metrics look stro…
1 sources - 4 claims
Time series train/test splits must preserve chronological order. The article uses a 75% training and 25% test split. Shuffling time series data with `train_test_split` causes data leakage. Data leakage can make evaluation metrics look strong while failing in live trading.