Use train_and_evaluate() rather than learn_runner about training-data-analyst

[Bhavyabhushanyadav]

@lakshmanok this code https://github.com/GoogleCloudPlatform/training-data-analyst/blob/acb1ded9234ebc0ab2ae1f3163ff43d68083b14d/courses/machine_learning/deepdive/05_artandscience/d_customestimator.ipynb can not perform training on cloudml.

When you use train_and_evaluate function it throws error and training is not performed on cloud ML

I am able to run this code on local machine

I have attached the screenshot of the error

{
insertId: "1aci8fzg6ycetxm"
jsonPayload: {
created: 1521028818.325793
levelname: "ERROR"
lineno: 345
message: "2018-03-14 12:00:18.325605: W tensorflow/core/framework/op_kernel.cc:1192] Out of range: Attempted to repeat an empty dataset infinitely."
pathname: "/runcloudml.py"
}
labels: {
compute.googleapis.com/resource_id: "2790092095787191020"
compute.googleapis.com/resource_name: "cmle-training-master-29e0961593-0-btrpt"
compute.googleapis.com/zone: "us-central1-c"
ml.googleapis.com/job_id: "plutots_180314_115915"
ml.googleapis.com/job_id/log_area: "root"
ml.googleapis.com/task_name: "master-replica-0"
ml.googleapis.com/trial_id: ""
}
logName: "projects/newproject-186115/logs/master-replica-0"
receiveTimestamp: "2018-03-14T12:00:40.702801781Z"
resource: {
labels: {
job_id: "plutots_180314_115915"
project_id: "newproject-186115"
task_name: "master-replica-0"
}
type: "ml_job"
}
severity: "ERROR"
timestamp: "2018-03-14T12:00:18.325793027Z"
}

[lakshmanok]

Make sure that gs://timeseries2/train.csv and eval.csv exist and are readable by the Cloud ML Service account