09.01 - Customer Data Analytics (Very Basic)

A smaller Project of the initial Idea 09 Customer Data Analytics

Starting Point

Since the challenge 09 Customer Data Analytics was not accepted, we did together with another not accepted Challenge Owner a small very basic analytics. We tried to find correlations or patterns in the provided File.

Approach

We used the file provided (anonymized 4'000 data from customers) using BigML and tried to find correlations in the data. The most valuable attributes seemed to be "Amount of contacts to customer center", "creditworthiness" and "Invoice-Type". It turned out, that (unfortunately very obvious) people with high contact numbers to the customer center also have a bad creditworthiness (German: Bonity).

Secondly we found a correlation, that the creditworthiness also correlates to the Invoice-Type. E-Bill-Customers pay their bill very fast (Bonity-Number is low whilst for customers with printed bills, it is much more likely to not pay the bills on time.

Other correlations - like between the Amount of energy could not be found.

Conclusion

All in all, the dataset was not very valuable (since it had to be cut down for data protection reasons - we are not allowed to use it internally also). That was probably the reason, why the challenge was not accepted. Learning: We have to think about how we want to deal with DA in the future and how we can use it compliant to data protection.

Title

Title

Event finish

Now serving beer!

3 years ago ~ benjamin_sawicki

Edited

3 years ago ~ nicola_ruch

Joined the team

3 years ago ~ benjamin_sawicki

Edited

3 years ago ~ nicola_ruch

Joined the team

3 years ago ~ PeterJanes

Other than that, the data is too bad for further analitics. But more data we cannot provide (monopole). Challenge aborted...

3 years ago ~ nicola_ruch

Results were not new: High contacts result in a high chance of bad bonity. The best paying customers use e-Bill or auto-payment.

3 years ago ~ nicola_ruch

The most valuable attributes were the amount of contacts to customer center, the Bonity-Index and the Billing-Type.

3 years ago ~ nicola_ruch

The original file (anonymized customer data) was analyzed on BigML using random forest.

3 years ago ~ nicola_ruch

Since the original challenge 05 and 09 was not accepted, we sat together thinking of what we can do with the ressources we had.

3 years ago ~ nicola_ruch

Edited

3 years ago ~ nicola_ruch

Joined the team

3 years ago ~ nicola_ruch

Challenge shared
Tap here to review.

3 years ago ~ nicola_ruch

Start

 
Contributed 3 years ago by nicola_ruch for Energy Data Hackdays 2021
All attendees, sponsors, partners, volunteers and staff at our hackathon are required to agree with the Hack Code of Conduct. Organisers will enforce this code throughout the event. We expect cooperation from all participants to ensure a safe environment for everybody. For more details on how the event is run, see the Guidelines on our website.

Creative Commons LicenceThe contents of this website, unless otherwise stated, are licensed under a Creative Commons Attribution 4.0 International License.