Channel estimation is a key functionality for millimeter wave (mmWave) MIMO communication systems. Channel state information is needed to design the precoders and combiners under different metrics such as mutual information or signal-to-interference-noise (SINR) ratio. Hybrid architectures are a commonly accepted solution for MIMO systems at mmWave, since they provide an interesting trade-off power consumption-achievable rate. Channel estimation is challenging however when using these architectures, since there is no direct access to the outputs of the different antenna elements in the array. The MIMO channel can only be observed through the analog combining network, which acts as a compression stage of the received signal. Most of prior work on channel estimation for hybrid architectures assumes a narrowband mmWave channel model. In this paper we consider a frequency selective mmWave channel and propose some compressive strategies to estimate the channel in the frequency domain. We evaluate the different algorithms and compute their complexity, so the trade-offs complexity-overhead-performance are clearly explained and compared to those of previous approaches.