A few years ago, during Bachelor final project, I and my colleague developed a system to represent and control domotic systems throught a personal computer or mobile device. The system was agnostic to domotic protocol and infrastructure dimensions.
Our system should be able to describe an apartment, a house, a factory or a city. The user could control is domotic system using mobile or dektop web interface.
To build the system we had the opportunity to explore several new technologies (at the time) like: .NET Micro Framework, ASP .NET 3.5, LINQ to Entities.
Was a very interesting challenge and pleasure!! If you want to know more about the project you can see Web Domus final presentation on slideshare or feel free to contact me.
Now I have a new challenge in hands, during this year I will be writing my master’s thesis on computer science. The thesis will be related to churn prediction on telecom companies.
During my research I found this study that shows some interesting approach into telecom churning prediction. The study is based on individual’s relationship extracted from Call Data Records (CDR’s).
Some of the conclusions are:
- “The trend implies that most prepaid customers call a relatively smaller number of people (friends)…”
- “The plot shows that most calls in the mobile network are short-lived”
- “caller’s tendency to finish a conversation within 60 seconds”
- The study find the churning probability grows with the number of “friends” that “churned” on the last month and even bigger when this pattern is recurrent on few months.
I will try to update the blog with my research on churning prediction.