Feature scaling is the process of normalizing or standardizing the range of independent variables (features) in data. Common methods include min-max scaling (rescaling features to [0,1] range) and standardization (subtract mean and divide by standard deviation, yielding zero mean and unit variance for each feature). Scaling is important for algorithms that are sensitive to feature magnitudes: for instance, distance-based methods (k-NN, clustering), or gradient-based optimization in neural networks/logistic regression where unscaled features can slow convergence. In SVMs or regularized models, if features aren’t scaled, the feature with larger values can dominate the objective function. One must fit the scaling on the training set and apply consistently to train/test to avoid data leakage. Some algorithms (like tree-based models) don’t need feature scaling because they are scale-invariant (splits are based on relative ordering).