The classical data-driven Takagi-Sugeno-Kang(TSK) fuzzy system extracts more features for structuring the antecedent of the fuzzy rule when trained by high dimensional data, and the interpretation of system is degenerated and the linguistic interpretation is complex. A fuzzy modeling model for the fuzzy subspace clustering based zero-order ridge regression TSK fuzzy system is proposed, in which the feature extraction mechanism based on the subspace feature of fuzzy subspace clustering is added, and the ridge regression is used to realize the learning of consequent. The proposed method not only can extract important features for structuring fuzzy rules, but also can extract different features for different rules. The experimental results on the synthetic and real-world datasets show the advantage of the proposed method.