Barbara Staron

Attribution Models in Marketing

/0 Comments/in , , , , , , , , , , , /by

Attribution Models

A Business and Statistical Case

INTRODUCTION

A desire to understand the causal effect of campaigns on KPIs

Advertising and marketing costs represent a huge and ever more growing part of the budget of companies. Studies have found out this share is as high as 10% and increases with the size of companies (CMO study by American Marketing Association and Duke University, 2017). Measuring precisely the impact of a specific marketing campaign on the sales of a company is a critical step towards an efficient allocation of this budget. Would the return be higher for an euro spent on a Facebook ad, or should we better spend it on a TV spot? How much should I spend on Twitter ads given the volume of sales this channel is responsible for?

Attribution Models have lately received great attention in Marketing departments to answer these issues. The transition from offline to online marketing methods has indeed permitted the collection of multiple individual data throughout the whole customer journey, and  allowed for the development of user-centric attribution models. In short, Attribution Models use the information provided by Tracking technologies such as Google Analytics or Webtrekk to understand customer journeys from the first click on a Facebook ad to the final purchase and adequately ponderate the different marketing campaigns encountered depending on their responsibility in the final conversion.

Issues on Causal Effects

A key question then becomes: how to declare a channel is responsible for a purchase? In other words, how can we isolate the causal effect or incremental value of a campaign ?

          1. A/B-Tests

One method to estimate the pure impact of a campaign is the design of randomized experiments, wherein a control and treated groups are compared.  A/B tests belong to this broad category of randomized methods. Provided the groups are a priori similar in every aspect except for the treatment received, all subsequent differences may be attributed solely to the treatment. This method is typically used in medical studies to assess the effect of a drug to cure a disease.

Main practical issues regarding Randomized Methods are:

  • Assuring that control and treated groups are really similar before treatment. Uually a random assignment (i.e assuring that on a relevant set of observable variables groups are similar) is realized;
  • Potential spillover-effects, i.e the possibility that the treatment has an impact on the non-treated group as well (Stable unit treatment Value Assumption, or SUTVA in Rubin’s framework);
  • The costs of conducting such an experiment, and especially the costs linked to the deliberate assignment of individuals to a group with potentially lower results;
  • The number of such experiments to design if multiple treatments have to be measured;
  • Difficulties taking into account the interaction effects between campaigns or the effect of spending levels. Indeed, usually A/B tests are led by cutting off temporarily one campaign entirely and measuring the subsequent impact on KPI’s compared to the situation where this campaign is maintained;
  • The dynamical reproduction of experiments if we assume that treatment effects may change over time.

In the marketing context, multiple campaigns must be tested in a dynamical way, and treatment effect is likely to be heterogeneous among customers, leading to practical issues in the lauching of A/B tests to approximate the incremental value of all campaigns. However, sites with a lot of traffic and conversions can highly benefit from A/B testing as it provides a scientific and straightforward way to approximate a causal impact. Leading companies such as Uber, Netflix or Airbnb rely on internal tools for A/B testing automation, which allow them to basically test any decision they are about to make.

References:

Books:

Experiment!: Website conversion rate optimization with A/B and multivariate testing, Colin McFarland, ©2013 | New Riders  

A/B testing: the most powerful way to turn clicks into customers. Dan Siroker, Pete Koomen; Wiley, 2013.

Blogs:

https://eng.uber.com/xp

https://medium.com/airbnb-engineering/growing-our-host-community-with-online-marketing-9b2302299324

Study:

https://cmosurvey.org/wp-content/uploads/sites/15/2018/08/The_CMO_Survey-Results_by_Firm_and_Industry_Characteristics-Aug-2018.pdf

        2. Attribution models

Attribution Models do not demand to create an experimental setting. They take into account existing data and derive insights from the variability of customer journeys. One key difficulty is then to differentiate correlation and causality in the links observed between the exposition to campaigns and purchases. Indeed, selection effects may bias results as exposure to campaigns is usually dependant on user-characteristics and thus may not be necessarily independant from the customer’s baseline conversion probabilities. For example, customers purchasing from a discount price comparison website may be intrinsically different from customers buying from FB ad and this a priori difference may alone explain post-exposure differences in purchasing bahaviours. This intrinsic weakness must be remembered when interpreting Attribution Models results.

                          2.1 General Issues

The main issues regarding the implementation of Attribution Models are linked to

  • Causality and fallacious reasonning, as most models do not take into account the aforementionned selection biases.
  • Their difficult evaluation. Indeed, in almost all attribution models (except for those based on classification, where the accuracy of the model can be computed), the additionnal value brought by the use of a given attribution models cannot be evaluated using existing historical data. This additionnal value can only be approximated by analysing how the implementation of the conclusions of the attribution model have impacted a given KPI.
  • Tracking issues, leading to an uncorrect reconstruction of customer journeys
    • Cross-device journeys: cross-device issue arises from the use of different devices throughout the customer journeys, making it difficult to link datapoints. For example, if a customer searches for a product on his computer but later orders it on his mobile, the AM would then mistakenly consider it an order without prior campaign exposure. Though difficult to measure perfectly, the proportion of cross-device orders can approximate 20-30%.
    • Cookies destruction makes it difficult to track the customer his the whole journey. Both regulations and consumers’ rising concerns about data privacy issues mitigate the reliability and use of cookies.1 – From 2002 on, the EU has enacted directives concerning privacy regulation and the extended use of cookies for commercial targeting purposes, which have highly impacted marketing strategies, such as the ‘Privacy and Electronic Communications Directive’ (2002/58/EC). A research was conducted and found out that the adoption of this ‘Privacy Directive’ had led to 64% decrease in advertising methods compared to the rest of the world (Goldfarb et Tucker (2011)). The effect was stronger for generalized sites (Yahoo) than for specialized sites.2 – Users have grown more and more conscious of data privacy issues and have adopted protective measures concerning data privacy, such as automatic destruction of cookies after a session is ended, or simply giving away less personnal information (Goldfarb et Tucker (2012) ) .Valuable user information may be lost, though tracking technologies evolution have permitted to maintain tracking by other means. This issue may be particularly important in countries highly concerned with data privacy issues such as Germany.
    • Offline/Online bridge: an Attribution Model should take into account all campaigns to draw valuable insights. However, the exposure to offline campaigns (TV, newspapers) are difficult to track at the user level. One idea to tackle this issue would be to estimate the proportion of conversions led by offline campaigns through AB testing and deduce this proportion from the credit assigned to the online campaigns accounted for in the Attribution Model.
    • Touch point information available: clicks are easy to follow but irrelevant to take into account the influence of purely visual campaigns such as display ads or video.

                          2.2 Today’s main practices

Two main families of Attribution Models exist:

  • Rule-Based Attribution Models, which have been used for in the last decade but from which companies are gradualy switching.

Attribution depends on the individual journeys that have led to a purchase and is solely based on the rank of the campaign in the journey. Some models focus on a single touch points (First Click, Last Click) while others account for multi-touch journeys (Bathtube, Linear). It can be calculated at the customer level and thus doesn’t require large amounts of data points. We can distinguish two sub-groups of rule-based Attribution Models:

  • One Touch Attribution Models attribute all credit to a single touch point. The First-Click model attributes all credit for a converion to the first touch point of the customer journey; last touch attributes all credit to the last campaign.
  • Multi-touch Rule-Based Attribution Models incorporate information on the whole customer journey are thus an improvement compared to one touch models. To this family belong Linear model where credit is split equally between all channels, Bathtube model where 40% of credit is given to first and last clicks and the remaining 20% is distributed equally between the middle channels, or time-decay models where credit assigned to a click diminishes as the time between the click and the order increases..

The main advantages of rule-based models is their simplicity and cost effectiveness. The main problems are:

– They are a priori known and can thus lead to optimization strategies from competitors
– They do not take into account aggregate intelligence on customer journeys and actual incremental values.
– They tend to bias (depending on the model chosen) channels that are over-represented at the beggining or end of the funnel, according to theoretical assumptions that have no observationnal back-ups.

  • Data-Driven Attribution Models

These models take into account the weaknesses of rule-based models and make a relevant use of available data. Being data-driven, following attribution models cannot be computed using single user level data. On the contrary values are calculated through data aggregation and thus require a certain volume of customer journey information.

References:

https://dspace.mit.edu/handle/1721.1/64920

 

        3. Data-Driven Attribution Models in practice

                          3.1 Issues

Several issues arise in the computation of campaigns individual impact on a given KPI within a data-driven model.

  • Selection biases: Exposure to certain types of advertisement is usually highly correlated to non-observable variables which are in turn correlated to consumption practices. Differences in the behaviour of users exposed to different campaigns may thus only be driven by core differences in conversion probabilities between groups whether than by the campaign effect.
  • Complementarity: it may be that campaigns A and B only have an effect when combined, so that measuring their individual impact would lead to misleading conclusions. The model could then try to assess the effect of combinations of campaigns on top of the effect of individual campaigns. As the number of possible non-ordered combinations of k campaigns is 2k, it becomes clear that inclusing all possible combinations would however be time-consuming.
  • Order-sensitivity: The effect of a campaign A may depend on the place where it appears in the customer journey, meaning the rank of a campaign and not merely its presence could be accounted for in the model.
  • Relative Order-sensitivity: it may be that campaigns A and B only have an effect when one is exposed to campaign A before campaign B. If so, it could be useful to assess the effect of given combinations of campaigns as well. And this for all campaigns, leading to tremendous numbers of possible combinations.
  • All previous phenomenon may be present, increasing even more the potential complexity of a comprehensive Attribution Model. The number of all possible ordered combination of k campaigns is indeed :

 

                          3.2 Main models

                                  A) Logistic Regression and Classification models

If non converting journeys are available, Attribition Model can be shaped as a simple classification issue. Campaign types or campaigns combination and volume of campaign types can be included in the model along with customer or time variables. As we are interested in inference (on campaigns effect) whether than prediction, a parametric model should be used, such as Logistic Regression. Non paramatric models such as Random Forests or Neural Networks can also be used though the interpretation of campaigns value would be more difficult to derive from the model results.

A common pitfall is the usual issue of spurious correlations on one hand and the correct interpretation of coefficients in business terms.

An advantage if the possibility to evaluate the relevance of the model using common model validation methods to evaluate its predictive power (validation set \ AUC \pseudo R squared).

                                  B) Shapley Value

Theory

The Shapley Value is based on a Game Theory framework and is named after its creator, the Nobel Price Laureate Lloyd Shapley. Initially meant to calculate the marginal contribution of players in cooperative games, the model has received much attention in research and industry and has lately been applied to marketing issues. This model is typically used by Google Adords and other ad bidding vendors. Campaigns or marketing channels are in this model seen as compementary players looking forward to increasing a given KPI.
Contrarily to Logistic Regressions, it is a non-parametric model. Contrarily to Markov Chains, all results are built using existing journeys, and not simulated ones.

Channels are considered to enter the game sequentially under a certain joining order. Shapley value try to The Shapley value of channel i is the weighted sum of the marginal values that channel i adds to all possible coalitions that don’t contain channel i.
In other words, the main logic is to analyse the difference of gains when a channel i is added after a coalition Ck of k channels, k<=n. We then sum all the marginal contributions over all possible ordered combination Ck of all campaigns excluding i, with k<=n-1.

Subsets framework

A first an most usual way to compute the Shapley Vaue is to consider that when a channel enters coalition, its additionnal value is the same irrelevant of the order in which previous channels have appeared. In other words, journeys (A>B>C) and (B>A>C) trigger the same gains.
Shapley value is computed as the gains associated to adding a channel i to a subset of channels, weighted by the number of (ordered) sequences that the (unordered) subset represents, summed up on all possible subsets of the total set of campaigns where the channel i is not present.
The Shapley value of the channel ???????? is then:

where |S| is the number of campaigns of a coalition S and the sum extends over all subsets S that do not not contain channel j. ????(????)  is the value of the coalition S and ????(???? ∪ {????????})  the value of the coalition formed by adding ???????? to coalition S. ????(???? ∪ {????????}) − ????(????) is thus the marginal contribution of channel ???????? to the coalition S.

The formula can be rewritten and understood as:

This method is convenient when data on the gains of on all possible permutations of all unordered k subsets of the n campaigns are available. It is also more convenient if the order of campaigns prior to the introduction of a campaign is thought to have no impact.

Ordered sequences

Let us define ????((A>B)) as the value of the sequence A then B. What is we let ????((A>B)) be different from ????((B>A)) ?
This time we would need to sum over all possible permutation of the S campaigns present before  ???????? and the N-(S+1) campaigns after ????????. Doing so we will sum over all possible orderings (i.e all permutations of the n campaigns of the grand coalition containing all campaigns) and we can remove the permutation coefficient s!(p-s+1)!.

This method is convenient when the order of channels prior to and after the introduction of another channel is assumed to have an impact. It is also necessary to possess data for all possible permutations of all k subsets of the n campaigns, and not only on all (unordered) k-subsets of the n campaigns, k<=n. In other words, one must know the gains of A, B, C, A>B, B>A, etc. to compute the Shapley Value.

Differences between the two approaches

We simulate an ordered case where the value for each ordered sequence k for k<=3 is known. We compare it to the usual Shapley value calculated based on known gains of unordered subsets of campaigns. So as to compare relevant values, we have built the gains matrix so that the gains of a subset A, B i.e  ????({B,A}) is the average of the gains of ordered sequences made up with A and B (assuming the number of journeys where A>B equals the number of journeys where B>A, we have ????({B,A})=0.5( ????((A>B)) + ????((B>A)) ). We let the value of the grand coalition be different depending on the order of campaigns-keeping the constraints that it averages to the value used for the unordered case.

Note: mvA refers to the marginal value of A in a given sequence.
With traditionnal unordered coalitions:

With ordered sequences used to compute the marginal values:

 

We can see that the two approaches yield very different results. In the unordered case, the Shapley Value campaign C is the highest, culminating at 20, while A and B have the same Shapley Value mvA=mvB=15. In the ordered case, campaign A has the highest Shapley Value and all campaigns have different Shapley Values.

This example illustrates the inherent differences between the set and sequences approach to Shapley values. Real life data is more likely to resemble the ordered case as conversion probabilities may for any given set of campaigns be influenced by the order through which the campaigns appear.

Advantages

Shapley value has become popular in allocation problems in cooperative games because it is the unique allocation which satisfies different axioms:

  • Efficiency: Shaple Values of all channels add up to the total gains (here, orders) observed.
  • Symmetry: if channels A and B bring the same contribution to any coalition of campaigns, then their Shapley Value i sthe same
  • Null player: if a channel brings no additionnal gains to all coalitions, then its Shapley Value is zero
  • Strong monotony: the Shapley Value of a player increases weakly if all its marginal contributions increase weakly

These properties make the Shapley Value close to what we intuitively define as a fair attribution.

Issues

  • The Shapley Value is based on combinatory mathematics, and the number of possible coalitions and ordered sequences becomes huge when the number of campaigns increases.
  • If unordered, the Shapley Value assumes the contribution of campaign A is the same if followed by campaign B or by C.
  • If ordered, the number of combinations for which data must be available and sufficient is huge.
  • Channels rarely present or present in long journeys will be played down.
  • Generally, gains are supposed to grow with the number of players in the game. However, it is plausible that in the marketing context a journey with a high number of channels will not necessarily bring more orders than a journey with less channels involved.

References:

R package: GameTheoryAllocation

Article:
Zhao & al, 2018 “Shapley Value Methods for Attribution Modeling in Online Advertising “
https://link.springer.com/content/pdf/10.1007/s13278-017-0480-z.pdf
Courses: https://www.lamsade.dauphine.fr/~airiau/Teaching/CoopGames/2011/coopgames-7%5b8up%5d.pdf
Blogs: https://towardsdatascience.com/one-feature-attribution-method-to-supposedly-rule-them-all-shapley-values-f3e04534983d

                                  B) Markov Chains

Markov Chains are used to model random processes, i.e events that occur in a sequential manner and in such a way that the probability to move to a certain state only depends on the past steps. The number of previous steps that are taken into account to model the transition probability is called the memory parameter of the sequence, and for the model to have a solution must be comprised between 0 and 4. A Markov Chain process is thus defined entirely by its Transition Matrix and its initial vector (i.e the starting point of the process).

Markov Chains are applied in many scientific fields. Typically, they are used in weather forecasting, with the sequence of Sunny and Rainy days following a Markov Process of memory parameter 0, so that for each given day the probability that the next day will be rainy or sunny only depends on the weather of the current day. Other applications can be found in sociology to understand the dynamics of social classes intergenerational reproduction. To get more both mathematical and applied illustration, I recommend the reading of this course.

In the marketing context, Markov Chains are an interesting way to model the conversion funnel. To go from the from the Markov Model to the Attribution logic, we calculate the Removal Effect of each channel, i.e the difference in conversions that happen if the channel is removed. Please read below for an introduction to the methodology.

The first step in a Markov Chains Attribution Model is to build the transition matrix that captures the transition probabilities between the campaigns accross existing customer journeys. This Matrix is to be read as a “From state A to state B” table, from the left to the right. A first difficulty is finding the right memory parameter to use. A large memory parameter would allow to take more into account interraction effects within the conversion funnel but would lead to increased computationnal time, a non-readable transition matrix, and be more sensitive to noisy data. Please note that this transition matrix provides useful information on the conversion funnel and on the relationships between campaigns and can be used as such as an analytical tool. I suggest the clear and easily R code which can be found here or here.

Here is an illustration of a Markov Chain with memory Parameter of 0: the probability to go to a certain campaign B in the next step only depend on the campaign we are currently at:

The associated Transition Matrix is then (with null probabilities left as Blank):

The second step is  to compute the actual responsibility of a channel in total conversions. As mentionned above, the main philosophy to do so is to calculate the Removal Effect of each channel, i.e the changes in the number of conversions when a channel is entirely removed. All customer journeys which went through this channel are settled out to be unsuccessful. This calculation is done by applying the transition matrix with and without the removed channels to an initial vector that contains the number of desired simulations.

Building on our current example, we can then settle an initial vector with the desired number of simulations, e.g 10 000:

 

It is possible at this stage to add a constraint on the maximum number of times the matrix is applied to the data, i.e on the maximal number of campaigns a simulated journey is allowed to have.

Advantages

  • The dynamic journey is taken into account, as well as the transition between two states. The funnel is not assumed to be linear.
  • It is possile to build a conversion graph that maps the customer journey provides valuable insights.
  • It is possible to evaluate partly the accuracy of the Attribution Model based on Markov Chains. It is for example possible to see how well the transition matrix help predict the future by analysing the number of correct predictions at any given step over all sequences.

Disadvantages

  • It can be somewhat difficult to set the memory parameter. Complementarity effects between channels are not well taken into account if the memory is low, but a parameter too high will lead to over-sensitivity to noise in the data and be difficult to implement if customer journeys tend to have a number of campaigns below this memory parameter.
  • Long journeys with different channels involved will be overweighted, as they will count many times in the Removal Effect.  For example, if there are n-1 channels in the customer journey, this journey will be considered as failure for the n-1 channel-RE. If the volume effects (i.e the impact of the overall number of channels in a journey, irrelevant from their type° are important then results may be biased.

References:

R package: ChannelAttribution

Git:

https://github.com/MatCyt/Markov-Chain/blob/master/README.md

Course:

https://www.ssc.wisc.edu/~jmontgom/markovchains.pdf

Article:

“Mapping the Customer Journey: A Graph-Based Framework for Online Attribution Modeling”; Anderl, Eva and Becker, Ingo and Wangenheim, Florian V. and Schumann, Jan Hendrik, 2014. Available at SSRN: https://ssrn.com/abstract=2343077 or http://dx.doi.org/10.2139/ssrn.2343077

“Media Exposure through the Funnel: A Model of Multi-Stage Attribution”, Abhishek & al, 2012

“Multichannel Marketing Attribution Using Markov Chains”, Kakalejčík, L., Bucko, J., Resende, P.A.A. and Ferencova, M. Journal of Applied Management and Investments, Vol. 7 No. 1, pp. 49-60.  2018

Blogs:

https://analyzecore.com/2016/08/03/attribution-model-r-part-1

https://analyzecore.com/2016/08/03/attribution-model-r-part-2

                          3.3 To go further: Tackling selection biases with Quasi-Experiments

Exposure to certain types of advertisement is usually highly correlated to non-observable variables. Differences in the behaviour of users exposed to different campaigns may thus only be driven by core differences in converison probabilities between groups whether than by the campaign effect. These potential selection effects may bias the results obtained using historical data.

Quasi-Experiments can help correct this selection effect while still using available observationnal data.  These methods recreate the settings on a randomized setting. The goal is to come as close as possible to the ideal of comparing two populations that are identical in all respects except for the advertising exposure. However, populations might still differ with respect to some unobserved characteristics.

Common quasi-experimental methods used for instance in Public Policy Evaluation are:

  • Discontinuity Regressions
  • Matching Methods, such as Exact Matching,  Propensity-score matching or k-nearest neighbourghs.

References:

Article:

“Towards a digital Attribution Model: Measuring the impact of display advertising on online consumer behaviour”, Anindya Ghose & al, MIS Quarterly Vol. 40 No. 4, pp. 1-XX, 2016

https://pdfs.semanticscholar.org/4fa6/1c53f281fa63a9f0617fbd794d54911a2f84.pdf

        4. First Steps towards a Practical Implementation

Identify key points of interests

  • Identify the nature of touchpoints available: is the data based on clicks? If so, is there a way to complement the data with A/B tests to measure the influence of ads without clicks (display, video) ? For example, what happens to sales when display campaign is removed? Analysing this multiplier effect would give the overall responsibility of display on sales, to be deduced from current attribution values given to click-based channels. More interestingly, what is the impact of the removal of display campaign on the occurences of click-based campaigns ? This would give us an idea of the impact of display ads on the exposure to each other campaigns, which would help correct the attribution values more precisely at the campaign level.
  • Define the KPI to track. From a pure Marketing perspective, looking at purchases may be sufficient, but from a financial perspective looking at profits, though a bit more difficult to compute, may drive more interesting results.
  • Define a customer journey. It may seem obvious, but the notion needs to be clarified at first. Would it be defined by a time limit? If so, which one? Does it end when a conversion is observed? For example, if a customer makes 2 purchases, would the campaigns he’s been exposed to before the first order still be accounted for in the second order? If so, with a time decay?
  • Define the research framework: are we interested only in customer journeys which have led to conversions or in all journeys? Keep in mind that successful customer journeys are a non-representative sample of customer journeys. Models built on the analysis of biased samples may be conservative. Take an extreme example: 80% of customers who see campaign A buy the product, VS 1% for campaign B. However, campaign B exposure is great and 100 Million people see it VS only 1M for campaign A. An Attribution Model based on successful journeys will give higher credit to campaign B which is an auguable conclusion. Taking into account costs per campaign (in the case where costs are calculated by clicks) may of course tackle this issue partly, as campaign A could then exhibit higher returns, but a serious fallacious reasonning is at stake here.

Analyse the typical customer journey    

  • Performing a duration analysis on the data may help you improve the definition of the customer journey to be used by your organization. After which days are converison probabilities null? Should we consider the effect of campaigns disappears after x days without orders? For example, if 99% of orders are placed in the 30 days following a first click, it might be interesting to define the customer journey as a 30 days time frame following the first oder.
  • Look at the distribution of the number of campaigns in a typical journey. If you choose to calculate the effect of campaigns interraction in your Attribution Model, it may indeed help you determine the maximum number of campaigns to be included in a combination. Indeed, you may not need to assess the impact of channel combinations with above than 4 different channels if 95% of orders are placed after less then 4 campaigns.
  • Transition matrixes: what if a campaign A systematically leads to a campaign B? What happens if we remove A or B? These insights would give clues to ask precise questions for a latter AB test, for example to find out if there is complementarity between channels A and B – (implying none should be removed) or mere substitution (implying one can be given up).
  • If conversion rates are available: it can be interesting to perform a survival analysis i.e to analyse the likelihood of conversion based on duration since first click. This could help us excluse potential outliers or individuals who have very low conversion probabilities.

Summary

Attribution is a complex topic which will probably never be definitively solved. Indeed, a main issue is the difficulty, or even impossibility, to evaluate precisely the accuracy of the attribution model that we’ve built. Attribution Models should be seen as a good yet always improvable approximation of the incremental values of campaigns, and be presented with their intrinsinc limits and biases.

Moussa El-Zaarah

A common trap when it comes to sampling from a population that intrinsically includes outliers

/0 Comments/in , , , , , , , /by

I will discuss a common fallacy concerning the conclusions drawn from calculating a sample mean and a sample standard deviation and more importantly how to avoid it.

Suppose you draw a random sample x_1, x_2, … x_N of size N and compute the ordinary (arithmetic) sample mean  x_m and a sample standard deviation sd from it.  Now if (and only if) the (true) population mean µ (first moment) and population variance (second moment) obtained from the actual underlying PDF  are finite, the numbers x_m and sd make the usual sense otherwise they are misleading as will be shown by an example.

By the way: The common correlation coefficient will also be undefined (or in practice always point to zero) in the presence of infinite population variances. Hopefully I will create an article discussing this related fallacy in the near future where a suitable generalization to Lévy-stable variables will be proposed.

 Drawing a random sample from a heavy tailed distribution and discussing certain measures

As an example suppose you have a one dimensional random walker whose step length is distributed by a symmetric standard Cauchy distribution (Lorentz-profile) with heavy tails, i.e. an alpha-stable distribution with alpha being equal to one. The PDF of an individual independent step is given by p(x) = \frac{\pi^{-1}}{(1 + x^2)} , thus neither the first nor the second moment exist whereby the first exists and vanishes at least in the sense of a principal value due to symmetry.

Still let us generate N = 3000 (pseudo) standard Cauchy random numbers in R* to analyze the behavior of their sample mean and standard deviation sd as a function of the reduced sample size n \leq N.

*The R-code is shown at the end of the article.

Here are the piecewise sample mean (in blue) and standard deviation (in red) for the mentioned Cauchy sampling. We see that both the sample mean and sd include jumps and do not converge.

Especially the mean deviates relatively largely from zero even after 3000 observations. The sample sd has no target due to the population variance being infinite.

If the data is new and no prior distribution is known, computing the sample mean and sd will be misleading. Astonishingly enough the sample mean itself will have the (formally exact) same distribution as the single step length p(x). This means that the sample mean is also standard Cauchy distributed implying that with a different Cauchy sample one could have easily observed different sample means far of the presented values in blue.

What sense does it make to present the usual interval x_m \pm sd / \sqrt{N} in such a case? What to do?

The sample median, median absolute difference (mad) and Inter-Quantile-Range (IQR) are more appropriate to describe such a data set including outliers intrinsically. To make this plausible I present the following plot, whereby the median is shown in black, the mad in green and the IQR in orange.

This example shows that the median, mad and IQR converge quickly against their assumed values and contain no major jumps. These quantities do an obviously better job in describing the sample. Even in the presence of outliers they remain robust, whereby the mad converges more quickly than the IQR. Note that a standard Cauchy sample will contain half of its sample in the interval median \pm mad meaning that the IQR is twice the mad.

Drawing a random sample from a PDF that has finite moments

Just for comparison I also show the above quantities for a standard normal (pseudo) sample labeled with the same color as before as a counter example. In this case not only do both the sample mean and median but also the sd and mad converge towards their expected values (see plot below). Here all the quantities describe the data set properly and there is no trap since there are no intrinsic outliers. The sample mean itself follows a standard normal, so that the sd in deed makes sense and one could calculate a standard error \frac{sd}{\sqrt{N}} from it to present the usual stochastic confidence intervals for the sample mean.

A careful observation shows that in contrast to the Cauchy case here the sampled mean and sd converge more quickly than the sample median and the IQR. However still the sampled mad performs about as well as the sd. Again the mad is twice the IQR.

And here are the graphs of the prementioned quantities for a pseudo normal sample:

The take-home-message:

Just be careful when you observe outliers and calculate sample quantities right away, you might miss something. At best one carefully observes how the relevant quantities change with sample size as demonstrated in this article.

Such curves should become of broader interest in order to improve transparency in the Data Science process and reduce fallacies as well.

Thank you for reading.

P.S.: Feel free to play with the set random seed in the R-code below and observe how other quantities behave with rising sample size. Of course you can also try different PDFs at the beginning of the code. You can employ a Cauchy, Gaussian, uniform, exponential or Holtsmark (pseudo) random sample.

 

QUIZ: Which one of the recently mentioned random samples contains a trap** and why?

**in the context of this article

 

R-code used to generate the data and for producing plots:

 

#R-script for emphasizing convergence and divergence of sample means

####install and load relevant packages ####

#uncomment these lines if necessary
#install.packages(c('ggplot2',’stabledist’))
#library(ggplot2)
#library(stabledist)

#####drawing random samples #####

#Setting a random seed for being able to reproduce results  
set.seed(1234567)   
N= 2000     #sample size

#Choose a PDF from which a sample shall be drawn
#To do so (un)comment the respective lines of following code

data <- rcauchy(N)    # option1(default): standard Cauchy sampling

#data <- rnorm(N)     #option2: standard Gaussian sampling
                               
#data <- rexp(N)    # option3: standard exponential sampling

#data <- rstable(N,alpha=1.5,beta=0)  # option4: standard symmetric Holtsmark sampling

#data <- runif(N)              #option5: standard uniform sample

#####descriptive statistics####
#preparations/declarations

SUM = vector()
sd =vector()
mean = vector()
SQ =vector()
SQUARES = vector()
median = vector()
mad =vector()
quantiles = data.frame()
sem =vector()

#piecewise calculaion of descrptive quantities

for (k in 1:length(data)){              #mainloop
SUM[k] <- sum(data[1:k])            # sum of sample
mean[k] <- mean(data[1:k])          # arithmetic mean
sd[k] <- sd(data[1:k])              # standard deviation
sem[k] <- sd[k]/(sqrt(k))          #standard error of the sample mean (for finite variances)
mad[k] <- mad(data[1:k],const=1)   # median absolute deviation    

for (j in 1:5){
qq <- quantile(data[1:k],na.rm = T)
quantiles[k,j] <- qq[j]         #quantiles of sample
}
colnames(quantiles) <- c('min','Q1','median','Q3','max')

for (i in 1:length(data[1:k])){
SQUARES[i] <- data[i]*data[i]    
}
SQ[k] <- sum(SQUARES[1:k])    #sum of squares of random sample
}  #end of mainloop

#create table containing all relevant data
TABLE <-  as.data.frame(cbind(quantiles,mean,sd,SQ,SUM,sem))




#####plotting results###
x11()
print(ggplot(TABLE,aes(1:N,median))+
geom_point(size=.5)+xlab('sample size n')+ylab('sample median'))
x11()
print(ggplot(TABLE,aes(1:N,mad))+geom_point(size=.5,color ='green')+
xlab('sample size n')+ylab('sample median absolute difference'))
x11()
print(ggplot(TABLE,aes(1:N,sd))+geom_point(size=.5,color ='red')+
xlab('sample size n')+ylab('sample standard deviation'))
x11()
print(ggplot(TABLE,aes(1:N,mean))+geom_point(size=.5, color ='blue')+
xlab('sample size n')+ylab('sample mean'))
x11()
print(ggplot(TABLE,aes(1:N,Q3-Q1))+geom_point(size=.5, color ='blue')+
xlab('sample size n')+ylab('IQR'))

#uncomment the following lines of code to see further plots

#x11()
#print(ggplot(TABLE,aes(1:N,sem))+geom_point(size=.5)+
#xlab('sample size n')+ylab('sample sum of r.v.'))
#x11()
#print(ggplot(TABLE,aes(1:N,SUM))+geom_point(size=.5)+
#xlab('sample size n')+ylab('sample sum of r.v.'))
#x11()
#print(ggplot(TABLE,aes(1:N,SQ))+geom_point(size=.5)+
#xlab('sample size n')+ylab('sample sum of squares'))

 

Vishal Bhalla

Statistical Relational Learning – Part 2

/0 Comments/in , , , , , , , /by

In the first part of this series on “An Introduction to Statistical Relational Learning”, I touched upon the basic Machine Learning paradigms, some background and intuition of the concepts and concluded with how the MLN template looks like. In this blog, we will dive in to get an in depth knowledge on the MLN template; again with the help of sample examples. I would then conclude by highlighting the various toolkit available and some of its differentiating features.

MLN Template – explained

A Markov logic network can be thought of as a group of formulas incorporating first-order logic and also tied with a weight. But what exactly does this weight signify?

Weight Learning

According to the definition, it is the log odds between a world where F is true and a world where F is false,

and captures the marginal distribution of the corresponding predicate.

Each formula can be associated with some weight value, that is a positive or negative real number. The higher the value of weight, the stronger the constraint represented by the formula. In contrast to classical logic, all worlds (i.e., Herbrand Interpretations) are possible with a certain probability [1]. The main idea behind this is that the probability of a world increases as the number of formulas it violates decreases.

Markov logic networks with its probabilistic approach combined to logic posit that a world is less likely if it violates formulas unlike in pure logic where a world is false if it violates even a single formula. Consider the case when a formula with high weight i.e. more significance is violated implying that it is less likely in occurrence.

Another important concept during the first phase of Weight Learning while applying an MLN template is “Grounding”. Grounding means to replace each variable/function in predicate with constants from the domain.

Weight Learning – An Example

Note: All examples are highlighted in the Alchemy MLN format

Let us consider an example where we want to identify the relationship between 2 different types of verb-noun pairs i.e noun subject and direct object.

The input predicateFormula.mln file contains

  1. The predicates nsubj(verb, subject) and dobj(verb, object) and
  2. Formula of nsubj(+ver, +s) and dobj(+ver, +o)

These predicates or rules are to learn all possible SVO combinations i.e. what is the probability of a Subject-Verb-Object combination. The + sign ensures a cross product between the domains and learns all combinations. The training database consists of the nsubj and dobj tuples i.e. relations is the evidence used to learn the weights.

When we run the above command for this set of rules against the training evidence, we learn the weights as here:

Note that the formula is now grounded by all occurrences of nsubj and dobj tuples from the training database or evidence and the weights are attached to it at the start of each such combination.

But it should be noted that there is no network yet and this is just a set of weighted first-order logic formulas. The MLN template we created so far will generate Markov networks from all of our ground formulas. Internally, it is represented as a factor graph.where each ground formula is a factor and all the ground predicates found in the ground formula are linked to the factor.

Inference

The definition goes as follows:

Estimate probability distribution encoded by a graphical model, for a given data (or observation).

Out of the many Inference algorithms, the two major ones are MAP & Marginal Inference. For example, in a MAP Inference we find the most likely state of world given evidence, where y is the query and x is the evidence.

which is in turn equivalent to this formula.

Another is the Marginal Inference which computes the conditional probability of query predicates, given some evidence. Some advanced inference algorithms are Loopy Belief Propagation, Walk-SAT, MC-SAT, etc.

The probability of a world is given by the weighted sum of all true groundings of a formula i under an exponential function, divided by the partition function Z i.e. equivalent to the sum of the values of all possible assignments. The partition function acts a normalization constant to get the probability values between 0 and 1.

Inference – An Example

Let us draw inference on the the same example as earlier.

After learning the weights we run inference (with or without partial evidence) and query the relations of interest (nsubj here), to get inferred values.

Tool-kits

Let’s look at some of the MLN tool-kits at disposal to do learning and large scale inference. I have tried to make an assorted list of all tools here and tried to highlight some of its main features & problems.

For example, BUGS i.e. Bayesian Logic uses a Swift Compiler but is Not relational! ProbLog has a Python wrapper and is based on Horn clauses but has No Learning feature. These tools were invented in the initial days, much before the present day MLN looks like.

ProbCog developed at Technical University of Munich (TUM) & the AI Lab at Bremen covers not just MLN but also Bayesian Logic Networks (BLNs), Bayesian Networks & ProLog. In fact, it is now GUI based. Thebeast gives a shell to analyze & inspect model feature weights & missing features.

Alchemy from University of Washington (UoW) was the 1st First Order (FO) probabilistic logic toolkit. RockIt from University of Mannheim has an online & rest based interface and uses only Conjunctive Normal Forms (CNF) i.e. And-Or format in its formulas.

Tuffy scales this up by using a Relational Database Management System (RDBMS) whereas Felix allows Large Scale inference! Elementary makes use of secondary storage and Deep Dive is the current state of the art. All of these tools are part of the HAZY project group at Stanford University.

Lastly, LoMRF i.e. Logical Markov Random Field (MRF) is Scala based and has a feature to analyse different hypothesis by comparing the difference in .mln files!

 

Hope you enjoyed the read. The content starts from basic concepts and ends up highlighting key tools. In the final part of this 3 part blog series I would explain an application scenario and highlight the active research and industry players. Any feedback as a comment below or through a message is more than welcome!

Back to Part I – Statistical Relational Learning

Additional Links:

[1] Knowledge base files in Logical Markov Random Fields (LoMRF)

[2] (still) nothing clever Posts categorized “Machine Learning” – Markov Logic Networks

[3] A gentle introduction to statistical relational learning: maths, code, and examples

Vishal Bhalla

Statistical Relational Learning

/1 Comment/in , , , /by

An Introduction to Statistical Relational Learning – Part 1

Statistical Relational Learning (SRL) is an emerging field and one that is taking centre stage in the Data Science age. Big Data has been one of the primary reasons for the continued prominence of this relational learning approach given, the voluminous amount of data available now to learn interesting and unknown patterns from data. Moreover, the tools have also improved their processing prowess especially, in terms of scalability.

This introductory blog is a prelude on SRL and later on I would also touch base on more advanced topics, specifically Markov Logic Networks (MLN). To start off, let’s look at how SRL fits into one of the 5 different Machine Learning paradigms.

Five Machine Learning Paradigms

Lets look at the 5 Machine Learning Paradigms: Each of which is inspired by ideas from a different field!

  1. Connectionists as they are called and led by Geoffrey Hinton (University of Toronto & Google and one of the major names in the Deep Learning community) think that a learning algorithm should mimic the brain! After all it is the brain that does all the complex actions for us and, this idea stems from Neuroscience.
  2. Another group of Evolutionists whose leader is the late John Holland (from the University of Michigan) believed it is not the brain but evolution that was precedent and hence the master algorithm to build anything. And using this approach of having the fittest ones program the future they are currently building 3D prints of future robots.
  3. Another thought stems from Philosophy where Analogists like Douglas R. Hofstadter an American writer and author of popular and award winning book – Gödel, Escher, Bach: an Eternal Golden Braid believe that Analogy is the core of Cognition.
  4. Symbolists like Stephen Muggleton (Imperial College London) think Psychology is the base and by developing Rules in deductive reasoning they built Adam – a robot scientist at the University of Manchester!
  5. Lastly we have a school of thought which has its foundations rested on Statistics & Logic, which is the focal point of interest in this blog. This emerging field has started to gain prominence with the invention of Bayesian networks 2011 by Judea Pearl (University of California Los Angeles – UCLA) who was awarded with the Turing award (the highest award in Computer Science). Bayesians as they are called, are the most fanatical of the lot as they think everything can be represented by the Bayes theorem using hypothesis which can be updated based on new evidence.

SRL fits into the last paradigm of Statistics and Logic. As such it offers another alternative to the now booming Deep Learning approach inspired from Neuroscience.

Background

In many real world scenario and use cases, often the underlying data is assumed to be independent and identically distributed (i.i.d.). However, real world data is not and instead consists of many relationships. SRL as such attempts to represent, model, and learn in the relational domain!

There are 4 main Models in SRL

  1. Probabilistic Relational Models (PRM)
  2. Markov Logic Networks (MLN)
  3. Relational Dependency Networks (RDN)
  4. Bayesian Logic Programs (BLP)

It is difficult to cover all major models and hence the focus of this blog is only on the emerging field of Markov Logic Networks.
MLN is a powerful framework that combines statistics (i.e. it uses Markov Random Fields) and logical reasoning (first order logic).

 

markov-random-fields-first-order-logic

Academia

Some of the prominent names in academic and the research community in MLN include:

  1. Professor Pedro Domingos from the University of Washington is credited with introducing MLN in his paper from 2006. His group created the tool called Alchemy which was one of the first, First Order Logic tools.
  2. Another famous name – Professor Luc De Raedt from the AI group at University of Leuven in Belgium, and their team created the tool ProbLog which also has a Python Wrapper.
  3. HAZY Project (Stanford University) led by Prof. Christopher Ré from the InfoLab is doing active research in this field and Tuffy, Felix, Elementary, Deep Dive are some of the tools developed by them. More on it later!
  4. Talking about academia close by i.e. in Germany, Prof. Michael Beetz and his entire team moved from TUM to TU Bremen. Their group invented the tool – ProbCog
  5. At present, Prof. Volker Tresp from Ludwig Maximilians University (LMU), Munich & Dr. Matthias Nickles at Technical University of Munich (TUM) have research interests in SRL.

Theory & Formulation

A look at some background and theoretical concepts to understand MLN better.

A. Basics – Probabilistic Graphical Models (PGM)

The definition of a PGM goes as such:

A PGM encodes a joint p(x,y) or conditional p(y|x) probability distribution such that given some observations we are provided with a full probability distribution over all feasible solutions.

A PGM helps to encode relationships between a set of random variables. And it achieves this by making use of a graph! These graphs can be either be Directed or Undirected Graphs.

B. Markov Blanket

A Markov Blanket is a Directed Acyclic graph. It is a Bayesian network and as you can see the central node A highlighted in red is dependent on its parents and parents of descendents (moralization) by the circle drawn around it. Thus these nodes are the only knowledge needed to predict node A.

C. Markov Random Fields (MRF)

A MRF is an Undirected graphical model. Every node in an MRF satisfies the Local Markov property of Conditional Independence, i.e. a node is conditionally independent of another node, given its neighbours. And now relating it to Markov Blanket as explained previously, a Markov blanket for a node is simply its adjacent nodes!

Intuition

We now that Probability handles uncertainty whereas Logic handles complexity. So why not make use of both of them to model relationships in data that is both uncertain and complex. Markov Logic Networks (MLN) precisely does that for us!

MLN is composed of a set of pairs of  <w, F> where F is the formula (written in FO logic) and weights (real numbers identifying the strength of the constraint).

MLN basically provides a template to ground a Markov network. Grounding would be explained in detail in the next but one section on “Weight Learning”.

It can be defined as a Log linear model where probability of a world is given by the weighted sum of all true groundings of a formula i under an exponential function. It is then divided by Z which is termed as the partition function and used to normalize and get probability values between 0 and 1.

propability_of_a_world_x

The MLN Template

Rules or Predicates

The relation to be learned is expressed in FO logic. Some of the different possible FO logical connectives and quantifiers are And (^), Or (V), Implication (→), and many more. Plus, Formulas may contain one or more predicates, connected to each other with logical connectives and quantified symbols.

Evidence

Evidence represent known facts i.e. the ground predicates. Each fact is expressed with predicates that contain only constants from their corresponding domains.

Weight Learning

Discover the importance of relations based on grounded evidence.

Inference

Query relations, given partial evidence to infer a probabilistic estimate of the world.

More on Weight Learning and Inference in the next part of this series!

Hope you enjoyed the read. I have deliberately kept the content basic and a mix of non technical and technical so as to highlight first the key players and some background concepts and generate the reader’s interest in this topic, the technicalities of which can easily be read in the paper. Any feedback as a comment below or through a message are more than welcome!

Continue reading with Statistical Relational Learning – Part II.

References