Heya, first of all, thanks for your talk (I am the guy with the dopamin kick, if you remember :) ) ! I got a small problem with plotting the predicted oil prices. For me it looks like the predicted values are a very flat graph:

I figured out what the problem is: the test_set is an array with one scalar vector (dtype=object).
y_test meanwhile is an array with one float64 vector.
Plotting the object values seems to produce the wrong graph.

One solution would be following:
plt.plot(np.float64(test_set)[timesteps:len(y_test)], color = 'red', label = 'Real Crude Oil Prices')

this fixes it for me:

The other plots further down and at the beginning would need the same treatment.

Doing it in this cell is probably not the ideal location tho.

For reference, I am running:

Python: 3.6.3
Keras: 2.1.3
tensorflow: 1.4.0

Currently

def readDataFrameFromCloudant(database):
    cloudantdata=spark.read.load(database, "org.apache.bahir.cloudant")
    (...)

This results in exception Py4JJavaError: An error occurred while calling o274.load. : java.lang.ClassNotFoundException: Failed to find data source: org.apache.bahir.cloudant.

Expected

def readDataFrameFromCloudant(database):
    cloudantdata=spark.read.load(database, "com.cloudant.spark")

In the final cell in the notebook

from rklib import submit
key = "XbAMqtjdEeepUgo7OOVwng"
part = "LqPRQ"
email = "###_YOUR_CODE_GOES_HERE_###"
secret = "###_YOUR_CODE_GOES_HERE_###"
submit(email, secret, key, part)

The submit function defined in rklib.py is called with 4 arguments while it's defined to take 6. all_parts and data are missing

score(data_healthy_scaled),
score(data_broken_scaled)

with method call
def score(data):
return model.#### your code here ###(data)

How did model calculate score without providing the label (or, y_test), as model.evaluate require x_test and y_test.

Currently

import pixiedust
pixiedust.installPackage("https://github.com/romeokienzler/developerWorks/raw/master/coursera/spark-sql-cloudant_2.11-2.3.0-SNAPSHOT.jar")

Result

"Link not found"

To solve this prob., I found new link in github.

https://github.com/iambenn/coursera-1/spark-sql-cloudant_2.11-2.3.0-SNAPSHOT.jar

Link connection was good, but It occurs another error.

Error

'NoneType' object has no attribute 'strip'

Solution(?)

import pixiedust
#pixiedust.installPackage("https://github.com/romeokienzler/developerWorks/raw/master/coursera/spark-sql-cloudant_2.11-2.3.0-SNAPSHOT.jar")

It works good.

Actually, I can't find what is real problem. maybe jar file, or service env.

Thanks

Hi,

Working through your tutorial, it looks like IBM Cloud no longer offers Python 2 notebooks, You might want to update the notebooks downloaded as part of the tutorial to work with Python 3.

Thanks.