I reduced my data with PCA. Can I get back to the original feature space? How much information did I lose?
Answer 1: Use inverse_transform
inverse_transform maps the reduced data back to the original feature space. The result is an approximation, because the information in the dropped components is gone.
pca = PCA(n_components=10).fit(X_scaled)
X_reduced = pca.transform(X_scaled)
X_restored = pca.inverse_transform(X_reduced)
If you scaled the data before PCA, the restored values are still scaled. Use scaler.inverse_transform(X_restored) to return to the original units.
