Generally you should be able to install tensorflow from pip using the described here way: https://www.tensorflow.org/versions/r0.11/get_started/os_setup.html#pip-installation (note, it’s not pip install tensorflow — you need to pick the correct build for your system).
Then you can just install the rest of dependencies via pip install scipy scikit-learn
Note, that skflow is now merged into TensorFlow itself and become sub-package called tf.learn, accessible via from tensorflow.contrib import learn.
If you manage to install tensorflow and other dependencies via pip, you don’t need bazel — you should be able just run via python path/to/example.py
If you do have bazel configured and working, you can run examples that are part of tf.learn while your current directory is clone of tensorflow repository:
bazel run //tensorflow/examples/learn:<name of example> (pick one from tensorflow/examples/learn folder).
Hope this helps!