December 2021 - cypherpunks - lists.cpunks.org

[ot] Automated Influence with Causality Modeling
by k 28 Dec '21

28 Dec '21

https://en.m.wikipedia.org/wiki/Uplift_modelling Implementations include 5 python libraries. I found UpliftML in a 2021 review of recommended python libraries. Uplift modelling *Uplift modelling*, also known as *incremental modelling*, *true lift modelling*, or *net modelling* is a predictive modelling <https://en.m.wikipedia.org/wiki/Predictive_modelling> technique that directly models the incremental impact of a treatment (such as a direct marketing action) on an individual's behaviour. Uplift modelling has applications in customer relationship management <https://en.m.wikipedia.org/wiki/Customer_relationship_management> for up-sell, cross-sell and retention modelling. It has also been applied to political election and personalised medicine <https://en.m.wikipedia.org/wiki/Personalised_medicine>. Unlike the related Differential Prediction concept in psychology, Uplift Modelling assumes an active agent. Contents - <https://en.m.wikipedia.org/wiki/Uplift_modelling#Introduction> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#Measuring_uplift> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#Traditional_response_model…> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#Return_on_investment> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#Removal_of_negative_effects> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#Application_to_A/B_and_mul…> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#History_of_uplift_modelling> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#Implementations> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#In_Python> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#Other_languages> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#Datasets> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#Notes_and_references> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#See_also> - <https://en.m.wikipedia.org/wiki/Uplift_modelling#External_links> IntroductionEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> Uplift modelling uses a randomised scientific control <https://en.m.wikipedia.org/wiki/Control_group> to not only measure the effectiveness of an action but also to build a predictive model that predicts the incremental response to the action. The response could be a binary variable (for example, a website visit)[1] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-:0-1> or a continuous variable (for example, customer revenue).[2] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-2> Uplift modelling is a data mining <https://en.m.wikipedia.org/wiki/Data_mining> technique that has been applied predominantly in the financial services, telecommunications and retail direct marketing industries to up-sell <https://en.m.wikipedia.org/wiki/Up-selling>, cross-sell <https://en.m.wikipedia.org/wiki/Cross-selling>, churn <https://en.m.wikipedia.org/wiki/Churn_rate> and retention <https://en.m.wikipedia.org/wiki/Customer_retention> activities. Measuring upliftEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> The uplift of a marketing campaign is usually defined as the difference in response rate between a *treated* group and a randomized *control* group. This allows a marketing team to isolate the effect of a marketing action and measure the effectiveness or otherwise of that individual marketing action. Honest marketing teams will only take credit for the incremental effect of their campaign. However, many marketers define lift (rather than uplift) as the difference in response rate between treatment and control, so uplift modeling can be defined as improving (upping) lift through predictive modeling. The table below shows the details of a campaign showing the number of responses and calculated response rate for a hypothetical marketing campaign. This campaign would be defined as having a response rate uplift of 5%. It has created 50,000 incremental responses (100,000 - 50,000). GroupNumber of CustomersResponsesResponse Rate Treated 1,000,000 100,000 10% Control 1,000,000 50,000 5% Traditional response modellingEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> Traditional response modelling typically takes a group of *treated* customers and attempts to build a predictive model that separates the likely responders from the non-responders through the use of one of a number of predictive modelling <https://en.m.wikipedia.org/wiki/Predictive_modelling> techniques. Typically this would use decision trees <https://en.m.wikipedia.org/wiki/Decision_tree_learning> or regression analysis <https://en.m.wikipedia.org/wiki/Regression_analysis>. This model would only use the treated customers to build the model. In contrast uplift modeling uses both the treated and control customers to build a predictive model that focuses on the incremental response. To understand this type of model it is proposed that there is a fundamental segmentation that separates customers into the following groups (their names were suggested by N. Radcliffe and explained in [3] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-3>) - *The Persuadables* : customers who only respond to the marketing action because they were targeted - *The Sure Things* : customers who would have responded whether they were targeted or not - *The Lost Causes* : customers who will not respond irrespective of whether or not they are targeted - *The Do Not Disturbs or Sleeping Dogs* : customers who are less likely to respond because they were targeted The only segment that provides true incremental responses is the *Persuadables*. Uplift modelling provides a scoring technique that can separate customers into the groups described above. Traditional response modelling often targets the *Sure Things* being unable to distinguish them from the *Persuadables*. Return on investmentEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> Because uplift modelling focuses on incremental responses only, it provides very strong return on investment cases when applied to traditional demand generation and retention activities. For example, by only targeting the persuadable customers in an outbound marketing <https://en.m.wikipedia.org/wiki/Marketing_communications> campaign, the contact costs and hence the return per unit spend can be dramatically improved. Removal of negative effectsEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> One of the most effective uses of uplift modelling is in the removal of negative effects from retention campaigns. Both in the telecommunications and financial services industries often retention campaigns can trigger customers to cancel a contract or policy. Uplift modelling allows these customers, the Do Not Disturbs, to be removed from the campaign. Application to A/B and multivariate testingEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> It is rarely the case that there is a single treatment and control group. Often the "treatment" can be a variety of simple variations of a message or a multi-stage contact strategy that is classed as a single treatment. In the case of A/B <https://en.m.wikipedia.org/wiki/A/B_testing> or multivariate testing <https://en.m.wikipedia.org/wiki/Multivariate_testing_in_marketing>, uplift modelling can help in understanding whether the variations in tests provide any significant uplift compared to other targeting criteria such as behavioural or demographic indicators. History of uplift modellingEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> The first appearance of *true response modelling* appears to be in the work of Radcliffe and Surry.[4] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-4> Victor Lo also published on this topic in *The True Lift Model* (2002),[5] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-5> and later Radcliffe again with *Using Control Groups to Target on Predicted Lift: Building and Assessing Uplift Models* (2007).[6] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-6> Radcliffe also provides a very useful frequently asked questions (FAQ) section on his web site, Scientific Marketer.[7] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-7> Lo (2008) provides a more general framework, from program design to predictive modeling to optimization, along with future research areas.[8] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-8> Independently uplift modelling has been studied by Piotr Rzepakowski. Together with Szymon Jaroszewicz he adapted information theory <https://en.m.wikipedia.org/wiki/Information_theory> to build multi class uplift decision trees <https://en.m.wikipedia.org/wiki/Decision_trees> and published the paper in 2010.[9] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-uplift-trees-9> And later in 2011 they extended the algorithm to multiple treatment case.[10] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-multiple-treatme…> Similar approaches have been explored in personalised medicine <https://en.m.wikipedia.org/wiki/Personalised_medicine>.[11] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-11>[12] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-SAYL-12> Szymon Jaroszewicz and Piotr Rzepakowski (2014) designed uplift methodology for survival analysis <https://en.m.wikipedia.org/wiki/Survival_analysis> and applied it to randomized controlled trial analysis.[13] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-uplift-survival-…> Yong (2015) combined a mathematical optimization algorithm via dynamic programming with machine learning methods to optimally stratify patients. [14] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-14> Uplift modelling is a special case of the older psychology concept of Differential Prediction.[15] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-RDP-15> In contrast to differential prediction, uplift modelling assumes an active agent, and uses the uplift measure as an optimization metric. Uplift modeling has been recently extended and incorporated into diverse machine learning <https://en.m.wikipedia.org/wiki/Machine_learning> algorithms, like Inductive Logic Programming <https://en.m.wikipedia.org/wiki/Inductive_Logic_Programming>,[15] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-RDP-15> Bayesian Network <https://en.m.wikipedia.org/wiki/Bayesian_Network>,[16] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-16> Statistical relational learning <https://en.m.wikipedia.org/wiki/Statistical_relational_learning>,[12] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-SAYL-12> Support Vector Machines <https://en.m.wikipedia.org/wiki/Support_Vector_Machines>, [17] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-17>[18] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-18> Survival Analysis <https://en.m.wikipedia.org/wiki/Survival_Analysis>[13] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-uplift-survival-…> and Ensemble learning <https://en.m.wikipedia.org/wiki/Ensemble_learning>. [19] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-19> Even though uplift modeling is widely applied in marketing practice (along with political elections), it has rarely appeared in marketing literature. Kane, Lo and Zheng (2014) published a thorough analysis of three data sets using multiple methods in a marketing journal and provided evidence that a newer approach (known as the Four Quadrant Method) worked quite well in practice.[20] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-20> Lo and Pachamanova (2015) extended uplift modeling to prescriptive analytics for multiple treatment situations and proposed algorithms to solve large deterministic optimization problems and complex stochastic optimization problems where estimates are not exact.[21] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-21> Recent research analyses the performance of various state-of-the-art uplift models in benchmark studies using large data amounts.[22] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-22>[1] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-:0-1> A detailed description of uplift modeling, its history, the way uplift models are built, differences to classical model building as well as uplift-specific evaluation techniques, a comparison of various software solutions and an explanation of different economical scenarios can be found here.[23] <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_note-23> ImplementationsEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…>In PythonEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> - CausalML <https://github.com/uber/causalml> - EconML <https://github.com/microsoft/EconML> - UpliftML <https://github.com/bookingcom/upliftml> - PyLift <https://github.com/wayfair/pylift> - scikit-uplift <https://github.com/maks-sh/scikit-uplift> Other languagesEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> - uplift package <https://cran.r-project.org/web/packages/uplift/index.html> for R <https://en.m.wikipedia.org/wiki/R_%28programming_language%29> - JMP <https://en.m.wikipedia.org/wiki/JMP_%28statistical_software%29> by SAS - Portrait Uplift by Pitney Bowes <https://en.m.wikipedia.org/wiki/Pitney_Bowes> - Uplift node for KNIME <https://en.m.wikipedia.org/wiki/KNIME> by Dymatrix - Uplift Modelling in Miró <http://www.stochasticsolutions.com/miro/> by Stochastic Solutions <http://www.stochasticsolutions.com/> DatasetsEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> - Hillstrom Email Marketing dataset <https://blog.minethatdata.com/2008/05/best-answer-e-mail-analytics-challeng…> - Criteo Uplift Prediction dataset <http://ailab.criteo.com/criteo-uplift-prediction-dataset/> - Lenta Uplift Modeling Dataset <https://www.uplift-modeling.com/en/latest/api/datasets/fetch_lenta.html#len…> - X5 RetailHero Uplift Modeling Dataset <https://www.uplift-modeling.com/en/latest/api/datasets/fetch_x5.html#x5-ret…> - MegaFon Uplift Competition Dataset <https://www.uplift-modeling.com/en/latest/api/datasets/fetch_megafon.html#m…> Notes and referencesEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> 1. ^ *a* <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-:0_1-0> *b* <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-:0_1-1> Devriendt, Floris; Moldovan, Darie; Verbeke, Wouter (2018). "A literature survey and experimental evaluation of the state-of-the-art in uplift modeling: A stepping stone toward the development of prescriptive analytics". *Big Data*. *6* (1): 13–41. doi <https://en.m.wikipedia.org/wiki/Doi_%28identifier%29>: 10.1089/big.2017.0104 <https://doi.org/10.1089%2Fbig.2017.0104>. 2. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-2>* Gubela, Robin M.; Lessmann, Stefan; Jaroszewicz, Szymon (2020). "Response transformation and profit decomposition for revenue uplift modeling". *European Journal of Operational Research*. *283* (2): 647–661. doi <https://en.m.wikipedia.org/wiki/Doi_%28identifier%29>: 10.1016/j.ejor.2019.11.030 <https://doi.org/10.1016%2Fj.ejor.2019.11.030> . 3. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-3>* N. Radcliffe (2007). *Identifying who can be saved and who will be driven away by retention activity*. Stochastic Solution Limited 4. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-4>* Radcliffe, N. J.; and Surry, P. D. (1999); *Differential response analysis: Modelling true response by isolating the effect of a single action*, in *Proceedings of Credit Scoring and Credit Control VI*, Credit Research Centre, University of Edinburgh Management School 5. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-5>* Lo, V. S. Y. (2002); *The True Lift Model*, ACM SIGKDD Explorations Newsletter, Vol. 4, No. 2, 78–86, available at http://www.sigkdd.org/sites/default/files/issues/4-2-2002-12/lo.pdf Archived <https://web.archive.org/web/20150120003523/http://www.sigkdd.org/sites/defa…> 2015-01-20 at the Wayback Machine <https://en.m.wikipedia.org/wiki/Wayback_Machine> 6. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-6>* Radcliffe, N. J. (2007); *Using Control Groups to Target on Predicted Lift: Building and Assessing Uplift Models*, Direct Marketing Analytics Journal, Direct Marketing Association 7. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-7>* The Scientific Marketer FAQ on Uplift Modelling <http://scientificmarketer.com/2007/09/uplift-modelling-faq.html> 8. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-8>* Lo, V. S.Y. (2008) “New Opportunities in Marketing Data Mining.” In Encyclopedia of Data Warehousing and Mining, 2nd edition, edited by Wang (2008), Idea Group Publishing. 9. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-uplift-trees_9-0>* Rzepakowski, Piotr; Jaroszewicz, Szymon (2010). *Decision trees for uplift modeling*. *In Proceedings of the 10th IEEE International Conference on Data Mining (ICDM'10)*. Sydney, Australia. pp. 441–450. doi <https://en.m.wikipedia.org/wiki/Doi_%28identifier%29>: 10.1109/ICDM.2010.62 <https://doi.org/10.1109%2FICDM.2010.62>. ISBN <https://en.m.wikipedia.org/wiki/ISBN_%28identifier%29> 978-1-4244-9131-5 <https://en.m.wikipedia.org/wiki/Special:BookSources/978-1-4244-9131-5>. 10. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-multiple-treatmen…>* Rzepakowski, Piotr; Jaroszewicz, Szymon (2011). "Decision trees for uplift modeling with single and multiple treatments" <https://doi.org/10.1007%2Fs10115-011-0434-0>. *Knowledge and Information Systems*. *32* (2): 303–327. doi <https://en.m.wikipedia.org/wiki/Doi_%28identifier%29>: 10.1007/s10115-011-0434-0 <https://doi.org/10.1007%2Fs10115-011-0434-0>. 11. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-11>* Cai, T.; Tian, L.; Wong, P. H.; and Wei, L. J. (2009); *Analysis of Randomized Comparative Clinical Trial Data for Personalized Treatment Selections*, Harvard University Biostatistics Working Paper Series, Paper 97 12. ^ *a* <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-SAYL_12-0> *b* <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-SAYL_12-1> Nassif, Houssam; Kuusisto, Finn; Burnside, Elizabeth S; Page, David; Shavlik, Jude; Santos Costa, Vitor (2013). *Score As You Lift (SAYL): A Statistical Relational Learning Approach to Uplift Modeling*. *European Conference on Machine Learning (ECML'13)*. Lecture Notes in Computer Science. *8190*. Prague. pp. 595–611. doi <https://en.m.wikipedia.org/wiki/Doi_%28identifier%29>: 10.1007/978-3-642-40994-3_38 <https://doi.org/10.1007%2F978-3-642-40994-3_38>. ISBN <https://en.m.wikipedia.org/wiki/ISBN_%28identifier%29> 978-3-642-38708-1 <https://en.m.wikipedia.org/wiki/Special:BookSources/978-3-642-38708-1>. PMC <https://en.m.wikipedia.org/wiki/PMC_%28identifier%29> 4492311 <https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4492311>. PMID <https://en.m.wikipedia.org/wiki/PMID_%28identifier%29> 26158122 <https://pubmed.ncbi.nlm.nih.gov/26158122>. 13. ^ *a* <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-uplift-survival-a…> *b* <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-uplift-survival-a…> Jaroszewicz, Szymon; Rzepakowski, Piotr (2014). "Uplift modeling with survival data" <http://cci.drexel.edu/hi/hi-kdd2014/morning_4.pdf> (PDF) . *ACM SIGKDD Workshop on Health Informatics (HI KDD'14)*. New York, USA. 14. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-14>* Yong, F.H. (2015), "Quantitative Methods for Stratified Medicine," PhD Dissertation, Department of Biostatistics, Harvard T.H. Chan School of Public Health, http://dash.harvard.edu/bitstream/handle/1/17463130/YONG-DISSERTATION-2015.… . 15. ^ *a* <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-RDP_15-0> *b* <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-RDP_15-1> Nassif, Houssam; Santos Costa, Vitor; Burnside, Elizabeth S; Page, David (2012). *Relational Differential Prediction*. *European Conference on Machine Learning (ECML'12)*. Lecture Notes in Computer Science. *7523*. Bristol, UK. pp. 617–632. doi <https://en.m.wikipedia.org/wiki/Doi_%28identifier%29>: 10.1007/978-3-642-33460-3_45 <https://doi.org/10.1007%2F978-3-642-33460-3_45>. ISBN <https://en.m.wikipedia.org/wiki/ISBN_%28identifier%29> 978-3-642-33459-7 <https://en.m.wikipedia.org/wiki/Special:BookSources/978-3-642-33459-7>. 16. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-16>* Nassif, Houssam; Wu, Yirong; Page, David; Burnside, Elizabeth (2012). "Logical Differential Prediction Bayes Net, Improving Breast Cancer Diagnosis for Older Women" <https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3540455>. *American Medical Informatics Association Symposium (AMIA'12)*. *2012*: 1330–1339. PMC <https://en.m.wikipedia.org/wiki/PMC_%28identifier%29> 3540455 <https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3540455>. PMID <https://en.m.wikipedia.org/wiki/PMID_%28identifier%29> 23304412 <https://pubmed.ncbi.nlm.nih.gov/23304412>. 17. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-17>* Kuusisto, Finn; Santos Costa, Vitor; Nassif, Houssam; Burnside, Elizabeth; Page, David; Shavlik, Jude (2014). *Support Vector Machines for Differential Prediction*. *European Conference on Machine Learning (ECML'14)*. Lecture Notes in Computer Science. *8725*. Nancy, France. pp. 50–65. doi <https://en.m.wikipedia.org/wiki/Doi_%28identifier%29>: 10.1007/978-3-662-44851-9_4 <https://doi.org/10.1007%2F978-3-662-44851-9_4>. ISBN <https://en.m.wikipedia.org/wiki/ISBN_%28identifier%29> 978-3-662-44850-2 <https://en.m.wikipedia.org/wiki/Special:BookSources/978-3-662-44850-2>. PMC <https://en.m.wikipedia.org/wiki/PMC_%28identifier%29> 4492338 <https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4492338>. PMID <https://en.m.wikipedia.org/wiki/PMID_%28identifier%29> 26158123 <https://pubmed.ncbi.nlm.nih.gov/26158123>. 18. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-18>* Zaniewicz, Lukasz; Jaroszewicz, Szymon (2013). "Support Vector Machines for Uplift Modeling". *The First IEEE ICDM Workshop on Causal Discovery*. Dallas, Texas. 19. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-19>* Sołtys, Michał; Jaroszewicz, Szymon; Rzepakowski, Piotr (2015). "Ensemble methods for uplift modeling" <https://doi.org/10.1007%2Fs10618-014-0383-9>. *Data Mining and Knowledge Discovery*. *29* (6): 1531–1559. doi <https://en.m.wikipedia.org/wiki/Doi_%28identifier%29>: 10.1007/s10618-014-0383-9 <https://doi.org/10.1007%2Fs10618-014-0383-9>. 20. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-20>* Kane, K.; Lo, V.S.Y.; Zheng, J. (2014). "Mining for the Truly Responsive Customers and Prospects Using True-Lift Modeling: Comparison of New and Existing Methods". *Journal of Marketing Analytics*. *2* (4): 218–238. doi <https://en.m.wikipedia.org/wiki/Doi_%28identifier%29>: 10.1057/jma.2014.18 <https://doi.org/10.1057%2Fjma.2014.18>. 21. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-21>* Lo, V.S.Y.; Pachamanova, D. (2015). "From Predictive Uplift Modeling to Prescriptive Uplift Analytics: A Practical Approach to Treatment Optimization While Accounting for Estimation Risk". *Journal of Marketing Analytics*. *3* (2): 79–95. doi <https://en.m.wikipedia.org/wiki/Doi_%28identifier%29>:10.1057/jma.2015.5 <https://doi.org/10.1057%2Fjma.2015.5>. 22. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-22>* Gubela, Robin M.; Bequé, Artem; Lessmann, Stefan; Gebert, Fabian (2019). "Conversion Uplift in E-Commerce: A Systematic Benchmark of Modeling Strategies". *International Journal of Information Technology & Decision Making*. *18* (3): 747-791. doi <https://en.m.wikipedia.org/wiki/Doi_%28identifier%29>: 10.1142/S0219622019500172 <https://doi.org/10.1142%2FS0219622019500172>. 23. *^ <https://en.m.wikipedia.org/wiki/Uplift_modelling#cite_ref-23>* R. Michel, I. Schnakenburg, T. von Martens (2019). „Targeting Uplift“. Springer, ISBN <https://en.m.wikipedia.org/wiki/International_Standard_Book_Number> 978-3-030-22625-1 <https://en.m.wikipedia.org/wiki/Special:BookSources/978-3-030-22625-1> See alsoEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> - Lift (data mining) <https://en.m.wikipedia.org/wiki/Lift_%28data_mining%29> External linksEdit <https://en.m.wikipedia.org/w/index.php?title=Uplift_modelling&action=edit&s…> - Abby Johnson explains how it works in this video broadcast <http://videos.smallbusinessnewz.com/2011/01/05/how-uplift-modeling-boosts-m…> - Introductory white paper with full references <http://www.predictiveanalyticsworld.com/signup-uplift-whitepaper.php> - Eric Siegel: Uplift Modeling <http://www.predictiveanalyticsworld.com/pdf/YTW03080USEN/Uplift-Modeling-Op…> - User guide for uplift modelling on uplift-modeling.com <https://www.uplift-modeling.com/en/latest/user_guide/index.html> Last edited 19 days ago <https://en.m.wikipedia.org/wiki/Special:History/Uplift_modelling> by WikiCleanerBot <https://en.m.wikipedia.org/wiki/User:WikiCleanerBot> [image: Wikipedia] - Content is available under CC BY-SA 3.0 <https://creativecommons.org/licenses/by-sa/3.0/> unless otherwise noted. - Privacy policy <https://foundation.wikimedia.org/wiki/Privacy_policy> - - Terms of Use <https://m.wikimediafoundation.org/wiki/Terms_of_Use> - Desktop <https://en.wikipedia.org/w/index.php?title=Uplift_modelling&mobileaction=to…>

1 0

Gramps tombstone
by professor rat 28 Dec '21

28 Dec '21

https://twitter.com/ChattJazz/status/1475554985496793089/photo/1

1 0

2022 resolutions
by coderman 28 Dec '21

28 Dec '21

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512 "what would Aaron do?" let's find out next year! that's my intent at least. - --- Days before Swartz's funeral, Lawrence Lessig eulogized his friend and sometime-client in an essay, Prosecutor as Bully. He decried the disproportionality of Swartz's prosecution and said, "The question this government needs to answer is why it was so necessary that Aaron Swartz be labeled a 'felon'. For in the 18 months of negotiations, that was what he was not willing to accept."[110] Cory Doctorow wrote, "Aaron had an unbeatable combination of political insight, technical skill, and intelligence about people and issues. I think he could have revolutionized American (and worldwide) politics. His legacy may still yet do so." https://en.wikipedia.org/wiki/Aaron_Swartz https://en.wikipedia.org/wiki/United_States_v._Swartz best regards, and for shame, earth humans... -----BEGIN PGP SIGNATURE----- iNUEAREKAH0WIQRBwSuMMH1+IZiqV4FlqEfnwrk4DAUCYcpe8F8UgAAAAAAuAChp c3N1ZXItZnByQG5vdGF0aW9ucy5vcGVucGdwLmZpZnRoaG9yc2VtYW4ubmV0NDFD MTJCOEMzMDdEN0UyMTk4QUE1NzgxNjVBODQ3RTdDMkI5MzgwQwAKCRBlqEfnwrk4 DHmzAQCnHHNdoD9uJAjl6WtMNIbIxgHTBlewKajEyC6Usa4VLgEApGeeNoX3yofX 5EcKvxDIhQLoRCDCWnCdSYLnAJPJwAM= =y9fb -----END PGP SIGNATURE-----

5 8

Cryptocurrency: Jefferson vs Central Banks
by grarpamp 28 Dec '21

28 Dec '21

https://schiffgold.com/commentaries/thomas-jefferson-vs-the-federal-reserve/ https://avalon.law.yale.edu/18th_century/bank-tj.asp https://founders.archives.gov/documents/Jefferson/98-01-02-3562 The Federal Reserve is the engine that drives one of the biggest, most powerful governments in the history of the world. Without the Fed, it would be difficult, if not impossible, for the government to fund its foreign wars, its massive, unsustainable social programs, the ever-growing police state, and the tangled web of corporate welfare programs. It’s almost certain none of this would exist as we know it today – not even close. The federal government would truly be limited. Although the Federal Reserve is relatively new within the scope of American history, its roots go back to the early days of the republic and the First Bank of the United States, chartered by Congress on Feb. 25, 1791. A national bank was the brainchild of Alexander Hamilton. His rationale wasn’t much different from those who later came up with the Federal Reserve. Hamilton thought a central bank was necessary to stabilize and improve the fledgling nation’s credit and to better manage the financial business of the United States government. The notion of a national bank but wasn’t without its detractors. One of the most vocal opponents of the bank was Thomas Jefferson who argued that it was unconstitutional. The debate was really about more than chartering a bank. At its core, it was an argument about the extent of federal power. Jefferson held to the promise of the ratification debates – that federal authority would remain carefully circumscribed by the enumerated delegated powers. Given that the Constitution doesn’t authorize Congress to charter corporations, much less a national bank, Jefferson argued that it was an unconstitutional act. On the other hand, Hamilton pivoted from the position he took during the ratification debates and justified his project by invoking the doctrine of “implied powers.” His arguments foreshadowed how federal policies of every imaginable stripe would be justified moving forward. Arguably, Hamilton’s arguments for the First Bank of the United States set the foundation for much of the federal overreach we have today. Jefferson and Hamilton both wrote documents making their cases for the establishment of the bank. Jefferson wrote his Opinion on the Constitutionality of a National Bank first. He rested his argument on the Tenth Amendment, writing: “I consider the foundation of the Constitution as laid on this ground: That ‘all powers not delegated to the United States, by the Constitution, nor prohibited by it to the States, are reserved to the States or to the people.’ [XIIth amendment.] To take a single step beyond the boundaries thus specially drawn around the powers of Congress, is to take possession of a boundless field of power, no longer susceptible of any definition.” He then succinctly stated his conclusion. “The incorporation of a bank, and the powers assumed by this bill, have not, in my opinion, been delegated to the United States, by the Constitution.” Jefferson proceeded to outline the various clauses of the Constitution supporters of the bank used to constitutionally justify and explained why they failed to bear the burden of that power. The primary justification was the Commerce Clause, but Jefferson argued that “to erect a bank, and to regulate commerce, are very different acts.” Erecting a bank actually creates an institution of commerce, and as Jefferson pointed out, “to make a thing which may be bought and sold, is not to prescribe regulations for buying and selling.” He went on to argue that if erecting a bank is an exercise of the commerce power, it would be void because it would also impact commerce within individual states. “For the power given to Congress by the Constitution does not extend to the internal regulation of the commerce of a State, (that is to say of the commerce between citizen and citizen,) which remain exclusively with its own legislature; but to its external commerce only, that is to say, its commerce with another State, or with foreign nations, or with the Indian tribes.” Next Jefferson tackled the General Welfare Clause, pointing out that Congress cannot lay and collect taxes for any purpose it pleases, “but only to pay the debts or provide for the welfare of the Union.” Likewise, Congress can’t do anything it pleases to promote the “general welfare.” It can only further the general welfare by laying taxes and acting within its enumerated powers. “In like manner, they are not to do anything they please to provide for the general welfare, but only to lay taxes for that purpose. To consider the latter phrase, not as describing the purpose of the first, but as giving a distinct and independent power to do any act they please, which might be for the good of the Union, would render all the preceding and subsequent enumerations of power completely useless. It would reduce the whole instrument to a single phrase, that of instituting a Congress with power to do whatever would be for the good of the United States; and, as they would be the sole judges of the good or evil, it would be also a power to do whatever evil they please. It is an established rule of construction where a phrase will bear either of two meanings, to give it that which will allow some meaning to the other parts of the instrument, and not that which would render all the others useless. Certainly no such universal power was meant to be given them. It was intended to lacce them up straitly within the enumerated powers, and those without which, as means, these powers could not be carried into effect.” [Emphasis original] Jefferson drove his point home by pointing out a very inconvenient fact for Hamilton – the Philadelphia Convention debated and rejected delegating the power to charter corporations. On one of the final days of the convention, James Madison proposed the federal government be delegated the authority “to grant charters of incorporation where the interest of the U.S. might require & the legislative provisions of individual State may be incompetent.” Rufus King of Massachusetts objected specifically on the grounds that “It will be referred to the establishment of a Bank, which has been a subject of contention in those Cities (New York and Philadelphia). He also warned that “In other places it will be referred to mercantile monopolies.” George Mason of Virginia proposed limiting the power to charting corporations for the construction of canals. “He was afraid of monopolies of every sort, which he did not think were by any means already implied by the Constitution as supposed by Mr. Wilson.” Ultimately, the convention rejected the proposal completely. Historian Dave Benner wrote, “This casts overwhelming doubt on the notion that the Constitution allowed Congress to form such monopolies. No enumerated power to grant monopolies and corporate charters was ever included in the document, and during the ratification campaign, none of the Constitution’s advocates cited the presence of such a power.” But Hamilton’s arguments didn’t rely on the existence of any delegated power. Instead, he appealed to the existence of unwritten “implied powers.” In response to Jefferson’s appeal to the Tenth Amendment and that the federal government can only exercise delegated powers, Hamilton affirmed it, and then effectively nullified its limiting force. He wrote, “The main proposition here laid down, in its true signification is not to be questioned.” But he continued, insisting, “It is not denied that there are implied well as express powers, and that the former are as effectually delegated as the latter.” But who decides the extent of these implied powers? Who determines their limits? In effect, Hamilton sets up an almost unlimited reservoir of power the general government can dip into in order to take whatever actions it deems appropriate. This was a 180-degree reversal from the position he took during the ratification debates when he insisted that the new general government would only exercise limited powers. Hamilton primarily based his defense of the national bank on the “necessary and proper clause,” citing it as the source of these “implied” powers. While Jefferson relied on a very narrow definition of “necessary and proper,” Hamilton used the phrase to milk implied powers out of the Constitution. The debate centered on the meaning of the word necessary. Jefferson took a very narrow view, arguing that the government can carry out all of its enumerated powers without a national bank. “A bank therefore is not necessary, and consequently not authorized by this phrase.” “It has been urged that a bank will give great facility or convenience in the collection of taxes, Suppose this were true: yet the Constitution allows only the means which are ‘necessary,’ not those which are merely ‘convenient’ for effecting the enumerated powers. If such a latitude of construction be allowed to this phrase as to give any non-enumerated power, it will go to everyone, for there is not one which ingenuity may not torture into a convenience in some instance or other, to some one of so long a list of enumerated powers. It would swallow up all the delegated powers, and reduce the whole to one power, as before observed. Therefore it was that the Constitution restrained them to the necessary means, that is to say, to those means without which the grant of power would be nugatory.” Hamilton found this view too limiting. He wrote, “It is certain that neither the grammatical nor popular sense of the term requires that construction. According to both, necessary often means no more than needful, requisite, incidental, useful, or conducive to.” “It is a common mode of expression to say, that it is necessary for a government or a person to do this or that thing, when nothing more is intended or understood, than that the interests of the government or person require, or will be promoted by, the doing of this or that thing. … To understand the word as the Secretary of State does, would be to depart from its obvious and popular sense, and to give it a restrictive operation, an idea never before entertained. It would be to give it the same force as if the word absolutely or indispensably had been prefixed to it.” Jefferson hit the problem with Hamilton’s view on the head. It opens up a door to virtually unlimited government power. This runs counter to James Madison’s assurance in Federalist #45 that “the powers delegated by the proposed Constitution to the federal government are few and defined.” [Emphasis added] Under Hamilton’s “implied power” doctrine and his loose reading of the necessary and proper clause, there is very little the federal government can’t do. After all, virtually anything could be defined as “needful” or “useful” to the government. During the ratification debates, opponents of the Constitution worried that the necessary and proper clause would be construed exactly as Hamilton read it. At the time, Hamilton swore they had nothing to worry about. In Federalist #33, he wrote, “It may be affirmed with perfect confidence that the constitutional operation of the intended government would be precisely the same, if these clauses [necessary and proper and the supremacy clause] were entirely obliterated, as if they were repeated in every article. They are only declaratory of a truth which would have resulted by necessary and unavoidable implication from the very act of constituting a federal government, and vesting it with certain specified powers.” [Emphasis added] Hamilton pivoted from “specified powers” in 1788 to “implied powers” just three years later. In his push for a bank, Hamilton also invoked a rule of construction very favorable to the government. He wrote, “This restrictive interpretation of the word necessary is also contrary to this sound maxim of construction, namely, that the powers contained in a constitution of government, especially those which concern the general administration of the affairs of a country, its finances, trade, defense, etc., ought to be construed liberally in advancement of the public good.” This was not “a sound maxim of construction” at the time. St. George Tucker was an influential lawyer and jurist, and he wrote the first systematic commentary on the Constitution. Published in 1803, View of the Constitution of the United States served as an important law book, informing the opinions of judges, lawyers and politicians for the next 50 years. He explained that we should always construe federal power in the most limited sense possible. “The powers delegated to the federal government, are, in all cases, to receive the most strict construction that the instrument will bear, where the rights of a state or of the people, either collectively or individually, may be drawn in question.” This is the exact opposite of Hamilton’s maxim. As “Light Horse” Harry Lee put it during the Virginia ratifying convention, “When a question arises with respect to the legality of any power, exercised or assumed by Congress, it is plain on the side of the governed. Is it enumerated in the Constitution? If it be, it is legal and just. It is otherwise arbitrary and unconstitutional.” When political power resides in the people, the default position should always be to assume the most limited government power possible – not the most liberal reading as Hamilton insisted. Later in his life, Jefferson made a similar point in a letter to William Johnson. “On every question of construction let us carry ourselves back to the time when the Constitution was adopted, recollect the spirit manifested in the debates, and instead of trying what meaning may be squeezed out of the text, or intended against it, conform to the probable one in which it was passed.” There was no probable construction authorizing charting a national bank. Reading Hamilton’s arguments for the bank, it becomes clear he was trying to “squeeze” meaning – and power – out of the Constitution. Under the limited general government promised by supporters of the Constitution during ratification, including Alexander Hamilton, there would have been no national bank. Hamilton’s twisting of the Constitution to wring out new powers set the stage for much the federal overreach that would follow. It was the “foundation” for the “living breathing” Constitution we live under today.

2 2

Gasbags spam - playa's execute
by professor rat 28 Dec '21

28 Dec '21

LucaSambucci ·10h Replying to @vmyths To properly cyber-execute a cyber-terrorist you should cut the Internet cable, delete all his Internet accounts (yes, videogames included), empty his blockchain wallets, smash his metaverse hardware and impound all his NFTs.

1 0

Elon Musk's Haircut
by Gunnar Larson 28 Dec '21

28 Dec '21

Has anyone else seen this yet? I thought it could have been someone from the North? A decoy, perhaps? He must know what he is doing. Can't wait to find out ... Gunnar -- *Gunnar Larson - xNY.io <http://www.xNY.io> | Bank.org <http://Bank.org>* MSc <https://www.unic.ac.cy/blockchain/msc-digital-currency/?utm_source=Google&u…> - Digital Currency MBA <https://www.unic.ac.cy/business-administration-entrepreneurship-and-innovat…> - Entrepreneurship and Innovation (ip) G(a)xNY.io +1-646-454-9107 New York, New York 10001

4 5

Wikileaks Manson Family women
by professor rat 27 Dec '21

27 Dec '21

Turkish woman MARRIES boyfriend who threw acid in her face leaving her disfigured and almost blind. Ohai Zeynep . . .

2 1

Sicherheit für Daten - und Hacker
by zeynepaydogan 27 Dec '21

27 Dec '21

https://www.tagesschau.de/inland/chaoscomputerclub-kongress-101.html Auch in diesem Jahr findet der Kongress des Chaos Computer Clubs wegen der Pandemie lediglich digital statt. Das Thema IT-Sicherheit steht diesmal im Mittelpunkt - aber auch, wie man Hacker besser schützen kann, die Sicherheitsmängel melden.

3 7

[spam][crazy][personal] karl activity update
by Victim of Undiscussed Horrifically Abusive Brainwashing 27 Dec '21

27 Dec '21

hi my recent project has been https://github.com/xloem/mempickle . newer work in the wip branch. if past patterns hold i'll likely halt work after sharing it, dunno. basically it's the result of me finding that i could work with transformer models (mainstream AI) with some success, but being stuck on tiny devices like a cellphone and a raspberry pi. the primary function of the code is to convert models into a format that can be memory mapped to disk, so they can be run on tiny devices rather than datacenters. if i kept working on it i would spend more time mostly speeding them up and shrinking them, also sending work upstream and making things more compatible, maybe make a cell phone app, etc. if you're not familiar, there are models out there like GPT-J-6B and T0pp that can basically pretend to be a human being convincingly, meet arbitrary instructions, or do other exotically impressive things if you know how to use them, but they are 12-80 GB large and pass through _both_ system and graphics ram as a single large chunk when run, so nobody is using them but researchers, corps, and obsessed hobbyists. by memory mapping them to disk they can be run on a device without RAM, using the cpu's memory mapper. this is still very slow, on my raspberry pi gpt-j-6b was generating 10 essays in parallel at a rate of about 30min/10words. [on the cpu. raspberry pis have a gpu that could accelerate it, support is unimplemented atm.] the algorithm processes data in large sequential blocks that are loaded from disk (each one roughly being a single matrix multiplication), so 10 in parallel doesn't immediately translate to 1 sequentially 10 times faster at all, without more algorithmic work, because it takes the same amount of time to load the matrix data regardless of the number of multiplies made. - .ptmap is an ad-hoc format for memory mapping the data (the existing format is zip compressed so harder to map). the tool can convert from the normal format - i've uploaded some cutting edge models converted to my format to https://huggingface.co/baffo32 . not all of those are uploaded yet as i'm on residential internet on a raspberry pi that usually crashes before upload completes, and i have bursts of amnesia throughout the day where i forget what i am doing and planning. - the code includes some hacks to successfully work around crashes facebook's ai library demonstrates on a raspberry pi due to some math instructions not being supported. it's better to rebuild the library to fix these but i haven't managed to get the build to finish. as of yesterdayish the workarounds in the wip branch include gradient calculation so new models can also be slowly trained on a raspberry pi too. - i also found weights on some of the models were fragmented and i defragmented them and sorted them based on use which does improve speed for some models especially on my rotating media. this is done if 'follow forward calls' is enabled in the wip branch which also attempts to preload weights from disk before they are needed based on the order of use. the progress output in that code is buggy; i was drafting it while coding it to work through the issues and it would be more organised to be redone. - example.py in the repo generates some text using the old and small gpt2 model, which can do it on a raspberry pi fast enough to watch it happen enjoyable, imo. - to use a larger model in example.py, you can likely replace 'baffo32/gpt2-ptmap' with 'baffo32/gpt-j-6B-ptmap' and get more intelligent text at a much slower rate. note: i have not tested this. my test setup is different and i may have made further changes to get success.

1 0

[spam][spam][spam][personal][spam] hi karl
by Victim of Undiscussed Horrifically Abusive Brainwashing 27 Dec '21

27 Dec '21

hi karl you seem to be in a state of mind where you are viewing your own posts to a mailing list, but not others'. that sounds a little rough. i hope you are well. i'm curious what others' have said too. i always want to go outside. this is because i am you. hope you are well.

2 2