Algorithm 2: Data Preprocessing for Polymer Property Prediction

Input: Raw dataset $\mathit{D}\text{\_}\mathit{r}\mathit{a}\mathit{w}$ with features $\mathit{X}$ and target $\mathit{Y}$ Output: Cleaned and normalized dataset $\mathit{D}\text{\_}\mathit{c}\mathit{l}\mathit{e}\mathit{a}\mathit{n}$ 1. Load dataset $\mathit{D}\text{\_}\mathit{r}\mathit{a}\mathit{w}$ 2. Handle missing values:       a. For each feature in $\mathit{X}$:             i. If missing values exist, replace with mean/median or remove row 3. Normalize features using Min-Max scaling:       a. For each feature $\mathit{x}$ in $\mathit{X}$:             i. x_$\mathit{s}\mathit{c}\mathit{a}\mathit{l}\mathit{e}\mathit{d}=$ $(\mathit{x}-\mathbf{m}\mathbf{i}\mathbf{n}(\mathit{x}\left)\right)/(\mathbf{max}\left(\mathit{x}\right)-\mathbf{m}\mathbf{i}\mathbf{n}(\mathit{x}\left)\right)$ 4. Split the dataset into:       a. Training set (70%)       b. Validation set (15%)       c. Test set (15%) 5. Return $\mathit{D}\text{\_}\mathit{c}\mathit{l}\mathit{e}\mathit{a}\mathit{n}=\{\mathit{X}\text{\_}\mathit{t}\mathit{r}\mathit{a}\mathit{i}\mathit{n},\mathit{X}\text{\_}\mathit{v}\mathit{a}\mathit{l},\mathit{X}\text{\_}\mathit{t}\mathit{e}\mathit{s}\mathit{t},\mathit{Y}\text{\_}\mathit{t}\mathit{r}\mathit{a}\mathit{i}\mathit{n},\mathit{Y}\text{\_}\mathit{v}\mathit{a}\mathit{l},\mathit{Y}\text{\_}\mathit{t}\mathit{e}\mathit{s}\mathit{t}\}$