My submission for the JKI Season 2 Challenge -1.Bit tied on wkdays so challenged the sunday to start the new season. while Used the time series and python component while in built nodes are more then sufficient.
Anil
3 Likes
Hello All,
This is my first post ever to the Just Knime It Challenges. (Although, I have been a Knime user for the last 2 years now). Happy to participate and see creative solutions from others as well. 
Best,
Pranab
Link to my hub page: Comparing Exchange Rates – KNIME Community Hub
3 Likes
At first, I also thought of using Python. But then, I challenged myself to do without it. Succeeded… somewhat!
Great plots, Anil.
3 Likes
Hello KNIMErs, Here is my solution for Challenge JKISeason2-1
#knime #JKISeason2-1 #dataanalysis #datascience
3 Likes
Here is my try to show correlation with respect to AUSTRUS-other currencies
3 Likes
Hi Everyone! My first post on the forum and what better way to do it than post my solution to Challenge 1!
My solution uses a -Single Selection Widget- node to decide whether to view the data as a Linear, Normalized or Logarithmic Line Plot with the -Line Plot (Labs)- node. The -Refresh Button Widget- node is used to update the data when choosing a different visualization:
I then use the -Linear Correlation- node and display the correlation matrix using the -Heat Map (Labs)- node:
The slider on the -Heat Map (Labs)- can be used to filter the matrix. For example, to display the most positively correlated currencies (Swiss Franc, Japanese Yen, Deutsche Mark and Dutch Guilder):
or the least correlated (for instance the Canadian Dollar and Deutsche Mark) :
The workflow is available from my HUB account at:
Take a look and I hope you enjoy it!
Heather
5 Likes
Congratulations Heather & welcome to the KNIME community forum !
Best
Ael
3 Likes
Hi,I posted my solution,Waiting for solution tomorrow.
1 Like
Hello knimers,
Here it is my first contribution to the JKI_S2’s challenges: Just KNIME It S02 _ CH01
Some features in my deployed solution are:
-
Statistical imputation based in week day differentials ( component )
-
Exchange rate tiles displaying Pearson’s and R^2 coefficients, for z-score normalized time series.
-
The dashboard can switch to display normalized values or normalized daily differentials.
BR
6 Likes
Hello, Everybody!
My firts post ever in Knime Hub, since I’m using it just a few months…
As far as the challenge, I tried to keep the solution quite simple, and included the answers in the workflow…
Hope to see all of you in the next challenges… 
Regards
Fabrício
3 Likes
my solution
2 Likes
2 Likes
@ujlayan , @egs, welcome to the KNIME forum!
Aiming to reflect on, that you have participated in the challenge; and to be considered for the Just KNIME It! Leaderboard as well:
“Remember to upload your solution with tag JKISeason2-1 ( link ) to your public space on the KNIME Community Hub.”
So, don’t forget to tag it!
BR
2 Likes
@niti.sethi , @arddashti , @lnunez02 , @elimisael
BR
P.S. a max of 5 mentions can be included in one post
1 Like
Just for giggles, I gave the challenge a go - perhaps spent too much time on this 
Couple of notes:
- The time period (1980-1998) covers the pre-Euro Exchange Rate Stability mechanism. Therefore, the Dutch Guilder, French France and Deutschmark should all be strongly correlated. There should also be periods in 1987 and 1991 when the British Pound was pegged to the Deutschmark and tracking the ERM which should also appear correlated - followed by periods of strong non-correlation.
- The workflow I used has two phases. The first of which is to ETL the data: read the data, convert the string date field to Java DateTime, rebase the dataset to include all days between the earliest and latest dates (makes charts easier to read), and convert the columns to ISO Currency codes (makes it easier to understand what the data relates to-more user focused).
- There are three mini-dashboard (more on those later). The first displays raw data; the other two show normalised data.
- The first approach to normalisation is to index all values to a fixed data in time (1980-01-02). Therefore the indexed values are ratio change from the base data. This allows a correlation over the entire time period to be calculated.
- The second approach is to normalise the data by calculating an index over a fixed time period (1, 7 or 28 days). This shows whether data is correlated over shorter time periods.
- The dashboard is a KNIME component. I used custom Python components because (a) they are easier to format exactly as I want (b) I can theme the charts to my liking (c) it gives me a bit more practice.
Workflow
Normalisation by indexing to base year
Normalisation over 28 days
It is possible using the Python Plotly components to zoom in on specific time periods, and to exclude traces to make chart clearer. The following is detail from the previous chart and shows how in 1982 the French Franc was having trouble maintaining correlation with the Dutch Guilder and Deutschmark.
Correlation plot (over the entire period, though should be possible to do correlations over short time periods).
5 Likes
Nice background information! @DiaAzul Thanks!
3 Likes
Hi folks,
Thank you so much for the incredible contributions! They were mindblowing! 
We just posted our solution to KNIME Hub.
We kept it simple, since understanding what currencies were the most correlated didn’t require too much effort. 
Still: so cool to see the composite visualizations that you folks came up with! Great practice for data apps!!
We hope to see you tomorrow for a new challenge! 
And remember to tag your solutions with JKISeason2-X (where X is the number of the challenge) before uploading them to KNIME Hub! This way they will count for our leaderboard. 
6 Likes
2 Likes
Glad to join this wonderful community, from Ecuador: My submission
]JKISeason2-1 Comparing Exchange Rates Jose De Souza.knwf (25.7 KB)
2 Likes