Size: 3745
Comment:
|
Size: 5267
Comment:
|
Deletions are marked like this. | Additions are marked like this. |
Line 12: | Line 12: |
=== Wednesday, 30.04.14 14.15 MAR4033 === * ''' Tillmann Fiehn''' (Chair: Mikio Braun) ==== Customer journey attribution with model based attributable risk ==== The efficiency of marketing investments is of high interest. However it is very difficult to be measured due to the split of marketing investments into multiple channels with different performance measures. Multi-touch attribution addresses performance based compensation to multiple channels with respect to the contribution to conversions. Click streams across multiple channels and conversion events, e.g. purchases, are called customer journeys. A logistic regression model that reflects both performance of touch points (e.g. clicks) and temporal relation to other touch points is fit to a sample of customer journeys. The probability estimates are used for the estimation of attributable risk for channels / a channel. That is:"the fraction of conversions that would not have been concluded if a channel / some channels did not serve ads". |
=== Wednesday, 04.06.14 14.15 MAR4033 === * ''' Guilhelm Simons (job talk)''' (Chair: Kristof Schütt) |
Line 27: | Line 18: |
=== Wednesday, 07.05.14 14.15 MAR4033 === * ''' Wiktor Pronobis''' (Chair: Guido Schwenk) === Wednesday, 14.05.14 14.15 MAR4033 === * ''' Sara Atito''' * ''' Forooz Shahbazi Avarvand''' * ''' Jan Saputra Müller''' (Chair: Sebastian Bach) === Wednesday, 21.05.14 14.15 MAR4033 === * ''' Jacquelyn Shelton''' * ''' Irene Winkler''' (Chair: Wiktor Pronobis) === Wednesday, 28.05.14 14.15 MAR4033 === * ''' Matthias Schultze-Kraft''' * ''' Danny Panknin''' (Chair: Felix Brockherde) === Wednesday, 04.06.14 14.15 MAR4033 === * ''' Alexander Bauer''' * ''' Kristof Schütt''' (Chair: Nico Görnitz) |
|
Line 55: | Line 21: |
* ''' Christian Hammerschmidt (job Talk)''' ''(Bi)graphical Models and Applications to Concurrency Theory'' Abstract: I will briefly introduce mobile process calculi for concurrent computing and focus on the Join calculus [1], which is readily distributed and quite similar to the actor model used in Scala/Akka. Bigraphs [2], a graphical meta model aiming to unify concurrent modelling, fails to encode Join calculi terms and their behaviour directly. In order to compare and reason about calculi within the bigraphical model, it needs to be extended. I will introduce the basic theory of bigraphs and give an encoding of Join into bigraphs by defining and applying orthogonal reactive systems. The encoding has a direct syntactic and semantic correspondence. The preservation of established bigraph theory will be outlined. Moreover, the concept of orthogonal systems can be used to define syntactical refinement operators as well as to embed computation layers into models. My work shows that orthogonal reactive systems are a conservative extension that allows the expression of a greater variety of features in bigraphs while retaining their established theory. [1] Fournet, C. and Gonthier, G. (1996) The Reflexive CHAM and the Join-calculus, Proceedings of the 23rd ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, POPL 1996 [2] Milner, R. (2008) Bigraphs and their algebra, Proceedings of the LIX Colloquium on Emerging Trends in Concurrency Theory (LIX 2006), ENTCS |
|
Line 59: | Line 51: |
* ''' Anne Porbadnigk''' | * ''' Owen Falzon (guest)''' |
Line 65: | Line 57: |
* ''' Jan Saputra Müller''' | |
Line 81: | Line 74: |
* ''' Anne Porbadnigk''' | |
Line 85: | Line 79: |
=== Wednesday, 28.05.14 14.15 MAR4033 === * ''' Danny Panknin''' * ''' Matthias Bussas (job talk)''' (Chair: Felix Brockherde) === Wednesday, 21.05.14 14.15 MAR4033 === * ''' Irene Winkler''' (Chair: Wiktor Pronobis) === Wednesday, 14.05.14 14.15 MAR4033 === * ''' Sara Atito''' * ''' Forooz Shahbazi Avarvand''' (Chair: Sebastian Bach) === Wednesday, 07.05.14 14.15 MAR4033 === * ''' Wiktor Pronobis''' (Chair: Guido Schwenk) === Wednesday, 30.04.14 14.15 MAR4033 === * ''' Tillmann Fiehn''' (Chair: Mikio Braun) ==== Customer journey attribution with model based attributable risk ==== The efficiency of marketing investments is of high interest. However it is very difficult to be measured due to the split of marketing investments into multiple channels with different performance measures. Multi-touch attribution addresses performance based compensation to multiple channels with respect to the contribution to conversions. Click streams across multiple channels and conversion events, e.g. purchases, are called customer journeys. A logistic regression model that reflects both performance of touch points (e.g. clicks) and temporal relation to other touch points is fit to a sample of customer journeys. The probability estimates are used for the estimation of attributable risk for channels / a channel. That is:"the fraction of conversions that would not have been concluded if a channel / some channels did not serve ads". |
Doktoranden-/Diplomandenseminar SS 14
In this seminar, master/phd students of the IDA group present their current work and research ideas.
When? |
usually on Wednesdays, 2:15pm to 3:15pm, or as announced (see below) |
Where? |
at TU Berlin , MAR 4005, or as announced (see below) |
How? |
More information for speakers and participants here. |
Next Talk
Wednesday, 04.06.14 14.15 MAR4033
Guilhelm Simons (job talk)
(Chair: Kristof Schütt)
Planned Talks
Wednesday, 11.06.14 14.15 MAR4033
Daniel Bartz
Bastian Venthur
Christian Hammerschmidt (job Talk)
(Bi)graphical Models and Applications to Concurrency Theory
Abstract: I will briefly introduce mobile process calculi for concurrent computing and focus on the Join calculus [1], which is readily distributed and quite similar to the actor model used in Scala/Akka. Bigraphs [2], a graphical meta model aiming to unify concurrent modelling, fails to encode Join calculi terms and their behaviour directly. In order to compare and reason about calculi within the bigraphical model, it needs to be extended. I will introduce the basic theory of bigraphs and give an encoding of Join into bigraphs by defining and applying orthogonal reactive systems. The encoding has a direct syntactic and semantic correspondence. The preservation of established bigraph theory will be outlined. Moreover, the concept of orthogonal systems can be used to define syntactical refinement operators as well as to embed computation layers into models. My work shows that orthogonal reactive systems are a conservative extension that allows the expression of a greater variety of features in bigraphs while retaining their established theory.
[1] Fournet, C. and Gonthier, G. (1996) The Reflexive CHAM and the Join-calculus, Proceedings of the 23rd ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, POPL 1996 [2] Milner, R. (2008) Bigraphs and their algebra, Proceedings of the LIX Colloquium on Emerging Trends in Concurrency Theory (LIX 2006), ENTCS (Chair: Matthias Schultze-Kraft)
Wednesday, 18.06.14 14.15 MAR4033
Fatemeh Ali Mardani
Owen Falzon (guest)
(Chair: Laura Acqualagna)
Wednesday, 25.06.14 14.15 MAR4033
Markus Wenzel
Sebastian Bach
Jan Saputra Müller
(Chair: Kristof Schütt)
Wednesday, 02.07.14 14.15 MAR4033
Laura Acqualagna
Johannes Höhne
(Chair: Daniel Bartz)
Wednesday, 09.07.14 14.15 MAR4033
Felix Brockherde
Irene Sturm
Sven Dähne
(Chair: Irene Winkler)
Wednesday, 16.07.14 14.15 MAR4033
Duncan Blythe
Nico Görnitz
Anne Porbadnigk
(Chair: Johannes Höhne)
Previous Talks
Wednesday, 28.05.14 14.15 MAR4033
Danny Panknin
Matthias Bussas (job talk)
(Chair: Felix Brockherde)
Wednesday, 21.05.14 14.15 MAR4033
Irene Winkler
(Chair: Wiktor Pronobis)
Wednesday, 14.05.14 14.15 MAR4033
Sara Atito
Forooz Shahbazi Avarvand
(Chair: Sebastian Bach)
Wednesday, 07.05.14 14.15 MAR4033
Wiktor Pronobis
(Chair: Guido Schwenk)
Wednesday, 30.04.14 14.15 MAR4033
Tillmann Fiehn
(Chair: Mikio Braun)
Customer journey attribution with model based attributable risk
The efficiency of marketing investments is of high interest. However it is very difficult to be measured due to the split of marketing investments into multiple channels with different performance measures. Multi-touch attribution addresses performance based compensation to multiple channels with respect to the contribution to conversions.
Click streams across multiple channels and conversion events, e.g. purchases, are called customer journeys. A logistic regression model that reflects both performance of touch points (e.g. clicks) and temporal relation to other touch points is fit to a sample of customer journeys. The probability estimates are used for the estimation of attributable risk for channels / a channel. That is:"the fraction of conversions that would not have been concluded if a channel / some channels did not serve ads".
-- Wojciech Samek -- Jul 2013