-
1. Re: A Web Data Connector using Python machine learning
Tom WDec 19, 2016 9:14 AM (in response to Erik Eklund)
Thanks for writing this up Erik. I'm curious if you've seen the release of the 'TabPy' integration for Python? Leverage the power of Python in Tableau with TabPy | Tableau Software
-
2. Re: A Web Data Connector using Python machine learning
Erik Eklund Dec 19, 2016 1:21 PM (in response to Tom W)1 of 1 people found this helpfulTabPy seems really nice. Fully integrated executable Python code from a workbook. Beautiful! Although, the WDC does have a interface for building predictive models, which could come in handy for those analyst not skilled in Python or R.
Thanks Tom for the info, appreciate it.
/Erik
-
3. Re: A Web Data Connector using Python machine learning
Prayson Wilfred Daniel May 23, 2017 6:16 AM (in response to Erik Eklund)This is brilliant. Not everyone in my team can Python.
BTW: I have some questions about data security issues. Is it possible for me to contact you via mail or LinkedIn?
-
4. Re: A Web Data Connector using Python machine learning
Tom WMay 23, 2017 7:30 AM (in response to Prayson Wilfred Daniel)
Prayson, your best bet is to submit your questions as a new thread on the forums. If you're looking for 1-1 support, your best bet is to probably engage some consulting / training services offered by Tableau.
-
5. Re: A Web Data Connector using Python machine learning
Prayson Wilfred Daniel May 23, 2017 7:55 AM (in response to Tom W)Thanks Tom. The data security issue is directed to Erik's Web Connector. When dealing with sensitive data, I want to know the how-s and where-s data is stored in pythonanywhere cloud and whether run this connection could be run in an internal sever.
-
6. Re: A Web Data Connector using Python machine learning
Erik Eklund May 24, 2017 3:45 PM (in response to Prayson Wilfred Daniel)1 of 1 people found this helpfulHi Prayson,
Thanks for reading and commenting on the WDC I built a while ago
I used my pythonanywhere account to host this WDC prototype as an example on how to connect and demonstrate the possibilities of a Web Data Connector using a ML library (scikit-learn) on the server side. It is not intended to handle sensitive data and I limited the file upload to a few MB.
However the Python code (using the web framework Flask) could preferably be placed and run from a local server or your local machine as long as it can handle http requests. I know when I was testing the WDC locally I connected Tableau to localhost:5000/wdc (5000 = the default port of Flask)
Yes it's true that not all business analysts and business users in need of predictive analytics are fluent in R or Python, but still they can have an excellent understanding of the business objective and the data thus providing valuable input variables for a model.
As demonstrated it works very well using a WDC with a ML library on the server providing predictive analytics in Tableau.
Although, I believe it would be even better if Tableau further developed the analytics capabilities and algorithms natively available in the desktop client. The k-means algorithm for clustering is a nice addition but additional algorithms in supervised learning would be much appreciated.
All the best
Erik
-
7. Re: A Web Data Connector using Python machine learning
Prayson Wilfred Daniel May 24, 2017 10:26 PM (in response to Erik Eklund)Thank you Erik. I have cloned it and will attempt to set it up on a local server.