It’s Humans, Not Algorithms, That Have a Bias Problem

Joshua New is a policy analyst at the Center for Data Innovation. Reposted from CDI’s blog.

Bias in big data. Automated discrimination. Algorithms that erode civil liberties.

These are some of the fears that the White House, the Federal Trade Commission, and other critics have expressed about an increasingly data-driven world. But these critics tend to forget that the world is already full of bias, and discrimination permeates human decision-making.

The truth is that the shift to a more data-driven world represents an unparalleled opportunity to crack down on unfair consumer discrimination by using data analysis to expose biases and reduce human prejudice. This opportunity is aptly demonstrated by the Consumer Financial Protection Bureau’s (CFPB) December 2013 auto loan discrimination suit against Ally Financial, the largest such suit in history, in which data and algorithms played a critical role in identifying and combating racial bias.

CFPB found that, from April 2011 to December 2013, Ally Financial had unfairly set higher interest rates on auto loans for 235,000 minority borrowers and ordered the company to pay out $80 million in damages. But the investigation also posed an interesting challenge: Since creditors are generally prohibited from collecting data on an applicant’s race, there was no hard evidence showing Ally had engaged in discriminatory practices. To piece together what really happened, CFPB used an algorithm to infer a borrower’s race based on other information in his or her loan application. Its analysis identified widespread overcharging of minority borrowers as a result of discriminatory interest rate markups at car dealerships.

Ally Financial buys retail installment contracts from more than 12,000 automobile dealers in the United States, essentially allowing dealers to act as middlemen for auto loans. If a consumer decides to finance his or her new car through a dealership rather than a bank, the dealership submits the consumer’s application to a company like Ally. If approved, the consumer pays back the dealership with interest. The interest rate, of course, matters a great deal. To determine what it will be, Ally calculates a “buy rate”—a minimum interest rate for which it is willing to purchase a retail installment contract, as determined by actuarial models. Ally notifies dealerships of this buy rate, but then also gives them substantial leeway to increase the interest rate to make the contract more profitable. Though consumers are free to negotiate these rates and shop around for the best deal, CFPB’s analysis determined that discretionary dealership pricing had a disparate impact on borrowers who were African American, Hispanic, Asian, or Pacific Islanders. On average, they paid between $200 and $300 more than similarly situated white borrowers.

Since creditors cannot inquire about race or ethnicity, Ally’s algorithmically generated buy rates are objective assessments. But when dealerships increase these rates, their judgments are entirely subjective, relying on humans to make decisions that could very well be influenced by racial bias. If dealerships instead took a similar approach to creditors and automated this decision-making process, there would be no opportunity for human bias to enter the equation. While dealerships could still increase interest rates to capture more profits, they could do so based on algorithmic analysis of predefined criteria about a consumer’s willingness to pay, thereby preventing themselves from offering similar consumers different rates based on their race.

Policymakers should guard against the possibility that automated decision-making could perpetuate bias, but with ever-increasing opportunities to collect and analyze data, the public and private sectors also should follow CFPB’s lead and identify new opportunities where data analytics can help expose and reduce human bias. For example, employers could rely on algorithms to select job applicants for interviews based on their objective qualifications rather than relying on human oversight that can be biased against factors such as whether or not the job applicant has an African American–sounding name. And taxi services could rely on algorithms to match drivers with riders rather than leaving it up to drivers who might be inclined to discriminate against passengers based on their race. If policymakers let fear of computerized decision-making impede wider deployment of fair algorithms, then society will lose a valuable opportunity to build a more just world.

Continua a leggere

Data for social good: suicide prevention

Earlier this week, GOOD Magazine published an interesting piece by Mark Hay on suicide prevention titled “Can Big Data Help Us Fight Rising Suicide Rates?” The part of the article that talks about data-driven prevention starts about halfway through. What follows is an excerpt from that section.

Yet there is one frontier in suicide prevention that seems especially promising, though in a way, it maybe a bit removed from the problem’s human element: big data predictions and intervention targeting.

We know that some populations are more likely than others to commit suicide. Men in the United States account for 79 percent of all suicides. People in their 20s are at higher risk than others. And whites and Native Americans tend to have higher suicide rates than other ethnicities. Yet we don’t have the greatest ability to grasp trends and other niche factors to build up actionable, targetable profiles of communities where we should focus our efforts. We’re stuck trying to expand a suicide prevention dragnet, as opposed to getting individuals at risk the precise information they need (even if they don’t tip off major signs to their friends and family).

That’s a big part of why last year, groups like the National Action Alliance for Suicide Prevention’s Research Prioritization Task Force listed better surveillance, data collection, and research on existing data as priorities for work in the field over the next decade. It’s also why multiple organizations are now developing algorithms to sort through diverse datasets, trying to identify behaviors, social media posting trends, language, lifestyle changes, or any other proxy that can help us predict suicidal tendencies. By doing this, the theory goes, we can target and deliver exactly the right information.

One of the greatest proponents of this data-heavy approach to suicide prevention is the United States Army, which suffers from a suicide rate many times higher than the general population. In 2012, they had more suicide deaths than casualties in Afghanistan. Yet with millions of soldiers stationed around the globe and limited suicide prevention resources, it’s been difficult to simply rely on expanding the dragnet. Instead, last December the Army announced that they’d developed an algorithm that distills the details of a soldier’s personal information into a set of 400 characteristics that mix and match to show whether an individual is likely in need of intervention. Their analysis isn’t perfect yet, but they’ve been able to identify a cluster of characteristics within 5 percent of military personnel who accounted for 52 percent of suicides, showing that they’re on the right track to better targeting and allocating prevention resources.

Yet perhaps the greatest distillation of this data-driven approach (combined with the expansive, barrier-reducing impulse of mainstream efforts) is the Crisis Text Line. Created in 2013 by organizers from DoSomething.org, the text line allows those too scared, embarrassed, or uncomfortable to vocalize their problems to friends, or over a hotline, to simply trace a pattern on a cell phone keypad (741741) and then type their problems in a text message. As of 2015, algorithmic learning allows the Crisis Text Line to search for keywords, based on over 8 million previous texts and data gathered from hundreds of suicide prevention workers, to identify who’s at serious risk and assign counselors to respond. But more than that, the data in texts can trip off time and vocabulary sensors, matching counselors with expertise in certain areas to respond to specific texters, or bringing up precisely tailored resources. For example, the system knows that self-harm peaks at 4 a.m. and that people typing “Mormon” are usually dealing with issues related to LGBTQ identity, discrimination, and isolation. Low-impact and low-cost with high potential for delivering the best information possible to those in need, it’s one of the cleverer young programs out there pushing the suicide prevention gains made over the last century.

It’ll be a few years before we can understand the impact of data analysis and targeting on suicide prevention efforts, especially relative to general attempts to expand existing programs. And given the limited success of a half-century of serious gains in understanding and resource provision, we’d be wise not to get our hopes up too much. But it’s not unreasonable to suspect that a combination of diversifying means of access, lowering barriers of communication, and better identifying those at risk could help us bring programs to populations that have not yet received them (or that we could not support quickly enough before). At the very least, crunching existing data may help us to discover why suicide rates have increased in recent years and to understand the mechanisms of this widespread social issue. We have solid, logical reason to support the development of programs like the Army’s algorithms and the Crisis Text Line, and to push for further similar initiatives. But really we have reason to support any kind of suicide prevention innovation, even if it feels less robust or promising than the recent data-driven efforts. If you’ve ever witnessed the pain that those moving towards suicide feel, or the wide-reaching fallout after someone takes his or her life, you’ll understand the visceral, human need to let a thousand flowers bloom, desperately hoping that one of them sticks. Hopefully, if data mining and targeting works well, that’ll only inspire further innovation, slowly putting a greater and greater dent in the phenomenon of suicide.

Continua a leggere

The reusable holdout: Preserving validity in adaptive data analysis

Moritz Hardt is a Research Scientist at Google. This post was originally published on the Google Research Blog.

Machine learning and statistical analysis play an important role at the forefront of scientific and technological progress. But with all data analysis, there is a danger that findings observed in a particular sample do not generalize to the underlying population from which the data were drawn. A popular XKCD cartoon illustrates that if you test sufficiently many different colors of jelly beans for correlation with acne, you will eventually find one color that correlates with acne at a p-value below the infamous 0.05 significance level.

Image credit: XKCD

Unfortunately, the problem of false discovery is even more delicate than the cartoon suggests. Correcting reported p-values for a fixed number of multiple tests is a fairly well understood topic in statistics. A simple approach is to multiply each p-value by the number of tests, but there are more sophisticated tools. However, almost all existing approaches to ensuring the validity of statistical inferences assume that the analyst performs a fixed procedure chosen before the data are examined. For example, “test all 20 flavors of jelly beans.” In practice, however, the analyst is informed by data exploration, as well as the results of previous analyses. How did the scientist choose to study acne and jelly beans in the first place? Often such choices are influenced by previous interactions with the same data. This adaptive behavior of the analyst leads to an increased risk of spurious discoveries that are neither prevented nor detected by standard approaches. Each adaptive choice the analyst makes multiplies the number of possible analyses that could possibly follow; it is often difficult or impossible to describe and analyze the exact experimental setup ahead of time.

In The Reusable Holdout: Preserving Validity in Adaptive Data Analysis, a joint work with Cynthia Dwork (Microsoft Research), Vitaly Feldman (IBM Almaden Research Center), Toniann Pitassi (University of Toronto), Omer Reingold (Samsung Research America) and Aaron Roth (University of Pennsylvania), to appear in Science tomorrow, we present a new methodology for navigating the challenges of adaptivity. A central application of our general approach is the reusable holdout mechanism that allows the analyst to safely validate the results of many adaptively chosen analyses without the need to collect costly fresh data each time.

The curse of adaptivity

A beautiful example of how false discovery arises as a result of adaptivity is Freedman’s paradox. Suppose that we want to build a model that explains “systolic blood pressure” in terms of hundreds of variables quantifying the intake of various kinds of food. In order to reduce the number of variables and simplify our task, we first select some promising looking variables, for example, those that have a positive correlation with the response variable (systolic blood pressure). We then fit a linear regression model on the selected variables. To measure the goodness of our model fit, we crank out a standard F-test from our favorite statistics textbook and report the resulting p-value.

Inference after selection: We first select a subset of the variables based on a data-dependent criterion and then fit a linear model on the selected variables.

Freedman showed that the reported p-value is highly misleading—even if the data were completely random with no correlation whatsoever between the response variable and the data points, we’d likely observe a significant p-value! The bias stems from the fact that we selected a subset of the variables adaptively based on the data, but we never account for this fact. There is a huge number of possible subsets of variables that we selected from. The mere fact that we chose one test over the other by peeking at the data creates a selection bias that invalidates the assumptions underlying the F-test.

Freedman’s paradox bears an important lesson. Significance levels of standard procedures do not capture the vast number of analyses one can choose to carry out or to omit. For this reason, adaptivity is one of the primary explanations of why research findings are frequently false as was argued by Gelman and Loken who aptly refer to adaptivity as “garden of the forking paths.”

Machine learning competitions and holdout sets

Adaptivity is not just an issue with p-values in the empirical sciences. It affects other domains of data science just as well. Machine learning competitions are a perfect example. Competitions have become an extremely popular format for solving prediction and classification problems of all sorts.

Each team in the competition has full access to a publicly available training set which they use to build a predictive model for a certain task such as image classification. Competitors can repeatedly submit a model and see how the model performs on a fixed holdout data set not available to them. The central component of any competition is the public leaderboard which ranks all teams according to the prediction accuracy of their best model so far on the holdout. Every time a team makes a submission they observe the score of their model on the same holdout data. This methodology is inspired by the classic holdout method for validating the performance of a predictive model.

Ideally, the holdout score gives an accurate estimate of the true performance of the model on the underlying distribution from which the data were drawn. However, this is only the case when the model is independent of the holdout data! In contrast, in a competition the model generally incorporates previously observed feedback from the holdout set. Competitors work adaptively and iteratively with the feedback they receive. An improved score for one submission might convince the team to tweak their current approach, while a lower score might cause them to try out a different strategy. But the moment a team modifies their model based on a previously observed holdout score, they create a dependency between the model and the holdout data that invalidates the assumption of the classic holdout method. As a result, competitors may begin to overfit to the holdout data that supports the leaderboard. This means that their score on the public leaderboard continues to improve, while the true performance of the model does not. In fact, unreliable leaderboards are a widely observed phenomenon in machine learning competitions.

Reusable holdout sets

A standard proposal for coping with adaptivity is simply to discourage it. In the empirical sciences, this proposal is known as pre-registration and requires the researcher to specify the exact experimental setup ahead of time. While possible in some simple cases, it is in general too restrictive as it runs counter to today’s complex data analysis workflows.

Rather than limiting the analyst, our approach provides means of reliably verifying the results of an arbitrary adaptive data analysis. The key tool for doing so is what we call the reusable holdout method. As with the classic holdout method discussed above, the analyst is given unfettered access to the training data. What changes is that there is a new algorithm in charge of evaluating statistics on the holdout set. This algorithm ensures that the holdout set maintains the essential guarantees of fresh data over the course of many estimation steps.

The limit of the method is determined by the size of the holdout set—the number of times that the holdout set may be used grows roughly as the square of the number of collected data points in the holdout, as our theory shows.

Armed with the reusable holdout, the analyst is free to explore the training data and verify tentative conclusions on the holdout set. It is now entirely safe to use any information provided by the holdout algorithm in the choice of new analyses to carry out, or the tweaking of existing models and parameters.

A general methodology

The reusable holdout is only one instance of a broader methodology that is, perhaps surprisingly, based on differential privacy—a notion of privacy preservation in data analysis. At its core, differential privacy is a notion of stability requiring that any single sample should not influence the outcome of the analysis significantly.

Example of a stable learning algorithm: Deletion of any single data point does not affect the accuracy of the classifier much.

A beautiful line of work in machine learning shows that various notions of stability imply generalization. That is any sample estimate computed by a stable algorithm (such as the prediction accuracy of a model on a sample) must be close to what we would observe on fresh data.

What sets differential privacy apart from other stability notions is that it is preserved by adaptive composition. Combining multiple algorithms that each preserve differential privacy yields a new algorithm that also satisfies differential privacy albeit at some quantitative loss in the stability guarantee. This is true even if the output of one algorithm influences the choice of the next. This strong adaptive composition property is what makes differential privacy an excellent stability notion for adaptive data analysis.

In a nutshell, the reusable holdout mechanism is simply this: access the holdout set only through a suitable differentially private algorithm. It is important to note, however, that the user does not need to understand differential privacy to use our method. The user interface of the reusable holdout is the same as that of the widely used classical method.

Reliable benchmarks

A closely related work with Avrim Blum dives deeper into the problem of maintaining a reliable leaderboard in machine learning competitions (see this blog post for more background). While the reusable holdout could directly be used for this purpose, it turns out that a variant of the reusable holdout, we call the Ladder algorithm, provides even better accuracy.

This method is not just useful for machine learning competitions, since there are many problems that are roughly equivalent to that of maintaining an accurate leaderboard in a competition. Consider, for example, a performance benchmark that a company uses to test improvements to a system internally before deploying them in a production system. As the benchmark data set is used repeatedly and adaptively for tasks such as model selection, hyper-parameter search and testing, there is a danger that eventually the benchmark becomes unreliable.

Conclusion

Modern data analysis is inherently an adaptive process. Attempts to limit what data scientists will do in practice are ill-fated. Instead we should create tools that respect the usual workflow of data science while at the same time increasing the reliability of data driven insights. It is our goal to continue exploring techniques that can help to create more reliable validation techniques and benchmarks that track true performance more accurately than existing methods.

Continua a leggere

Data for Good in Bangalore

Miriam Young is a Communications Specialist at DataKind.

At DataKind, we believe the same algorithms and computational techniques that help companies generate profit can help social change organizations increase their impact. As a global nonprofit, we harness the power of data science in the service of humanity by engaging data scientists and social change organizations on projects designed to address critical social issues.

Our global Chapter Network recently wrapped up a marathon of DataDives, helping local organizations with their data challenges over the course of a weekend. This post highlights two of the projects from DataKind Bangalore’s first DataDive earlier this year, where volunteers used data science to help support rural agriculture and combat urban corruption.

Digital Green

Founded in 2008, Digital Green is an international, nonprofit development organization that builds and deploys information and communication technology to amplify the effectiveness of development efforts to affect sustained social change. They have a series of educational videos of agricultural best practices to help farmers in villages succeed.

The Challenge

Help farmers more easily find videos relevant to them by developing a recommendation engine that suggests videos based on open data on local agricultural conditions. The team was working with a collection of videos, each focused on a specific crop, along with descriptions, but each description was in a different regional language. The challenge, then, was parsing and interpreting this information to use it as as a descriptive feature for the video. To add another challenge, they needed geodata with the geographical boundaries of different regions to map the videos to a region with specific soil types and environmental conditions, but the data didn’t exist.

The Solution

The volunteers got to work preparing this dataset and published boundaries of 103,344 indian villages and geocoded 1062 Digital Green villages in Madhya Pradesh(MP) to 22 soil polygons. They then clustered 22 MP districts based on 179 feature vectors. They also mapped the villages that Digital Green works with into 5 agro-climatic clusters. Finally, the team developed a Hinglish parser that parses the Hindi titles of available videos and translates them to English to help the recommender system understand which crop the videos relate to.

I Change My City / Janaagraha

Janaagraha was established in 2001 as a nonprofit that aims to combine the efforts of the government and citizens to ensure better quality of life in cities by improving urban infrastructure, services and civic engagement. Their civic portal, IChangeMyCity promotes civic action at a neighborhood level by enabling citizens to report a complaint that then gets upvoted by the community and flagged for government officials to take action.

The Challenge

Deal with duplicate complaints that can clog the system and identify factors that delay open issues from being closed out.

The Solution

To deal with the problem of duplicate complaints, the team used Jaccard similarity and Cosine similarity on vectorized complaints to cluster similar complaints together. Disambiguation was performed by ward and geography. The model they built delivered a precision of more than 90%.

To deal with the problem of identifying factors affecting closure by user and authorities, the team used two approaches. The first approach involved analysis using Decision Trees by capturing attributes like Comments, Vote-ups, Agency ID, Subcategory and so on. The second approach involved logistic regression to predict closure probability. Closure probability was modeled as a function of complaint subcategory, ward, comment velocity, vote-ups and similar other factors.

With these new features, iChangeMyCity will be able to better handle the large volume of incoming requests and Digital Green will be better able to serve farmers.

These initial findings are certainly valuable, but DataDives are actually much bigger than just weekend events. The weeks of preparation that go into them and months of impact that ripple out from them make them a step in an organization’s larger data science journey. This is certainly the case here, as both of these organizations are now exploring long-term projects with DataKind Bangalore to expand on this work.

Stay tuned for updates on these exciting projects to see what happens next!

Interested in getting involved? Find your local chapter and sign up to learn more about our upcoming events.

Continua a leggere

Exploring the world of data-driven innovation

Mike Masnick is founder of the Copia Institute.

In the last few years, there’s obviously been a tremendous explosion in the amount of data floating around. But we’ve also seen an explosion in the efforts to understand and make use of that data in valuable and important ways. The advances, both in terms of the type and amount of data available, combined with advances in computing power to analyze the data, are opening up entirely new fields of innovation that simply weren’t possible before.

We recently launched a new think tank, the Copia Institute, focused on looking at the big challenges and opportunities facing the innovation world today. An area we’re deeply interested in is data-driven innovation. To explore this space more thoroughly, the Copia Institute is putting together an ongoing series of case studies on data-driven innovation, with the first few now available in the Copia library.

Our first set of case studies includes a look at how the Polymerase Chain Reaction (PCR) helped jumpstart the biotechnology field today. PCR is, in short, a machine for copying DNA, something that was extremely difficult to do (outside of living things copying their own DNA). The discovery was something of an accident: A scientist discovered that certain microbes survived in the high temperatures of the hot springs of Yellowstone National Park, previously thought impossible. This resulted in further study that eventually led to the creation of PCR.

PCR was patented but licensed widely and generously. It basically became the key to biotech and genetic research in a variety of different areas. The Human Genome Project, for example, was possible only thanks to the widespread availability of PCR. Those involved in the early efforts around PCR were actively looking to share the information and concept rather than lock it up entirely, although there were debates about doing just that. By making sure that the process was widely available, it helped to accelerate innovation in the biotech and genetics fields. And with the recent expiration of the original PCR patents, the technology is even more widespread today, expanding its contribution to the field.

Another case study explores the value of the HeLa cells in medical research—cancer research in particular. While the initial discovery of HeLa cells may have come under dubious circumstances, their contribution to medical advancement cannot be overstated. The name of the HeLa cells comes from the patient they were originally taken from, a woman named Henrietta Lacks. Unlike previous human cell samples, HeLa cells continued to grow and thrive after being removed from Henrietta. The cells were made widely available and have contributed to a huge number of medical advancements, including work that has resulted in five Nobel prizes to date.

With both PCR and HeLa cells, we saw an important pattern: an early discovery that was shared widely, enabling much greater innovation to flow from proliferation of data. It was the widespread sharing of information and ideas that contributed to many of these key breakthroughs involving biotechnology and health.

At the same time, both cases raise certain questions about how to best handle similar developments in the future. There are questions about intellectual property, privacy, information sharing, trade secrecy and much more. At the Copia Institute, we plan to more dive into many of these issues with our continuing series of case studies, as well as through research and events.

Continua a leggere

DataEDGE: A New Vision for Data Science

Steven Weber is a professor in the School of Information and Political Science department at UC Berkeley.

It’s commonly said that most people overestimate the impact of technology in the short term, and underestimate its impact over the longer term.

Where is Big Data in 2013? Starting to get very real, in our view, and right on the cusp of underestimation in the long term. The short term hype cycle is (thankfully) burning itself out, and the profound changes that data science can and will bring to human life are just now coming into focus. It may be that Data Science is right now about where the Internet itself was in 1993 or so. That’s roughly when it became clear that the World Wide Web was a wind that would blow across just about every sector of the modern economy while transforming foundational things we thought were locked in about human relationships, politics, and social change. It’s becoming a reasonable bet that Data Science is set to do the same—again, and perhaps even more profoundly—over the next decade. Just possibly, more quickly than that.

There are important differences which have equally come into focus. Let’s face it: Data Science is just plain hard to do, in a way that the Web was not. Data is technically harder, from a hardware and a software perspective. It’s intellectually harder, because the expertise and disciplines needed to work with this kind of data span (at a minimum) computer science, statistics, mathematics, and—controversially—domain expertise in the area of application. And it will be harder to manage issues of ethics, privacy, and access, precisely because the data revolution is, well, really a revolution.

Can data, no matter how big, change the world for the better? It may be the case that in some fields of human endeavor and behavior, the scientific analysis of big data by itself will create such powerful insights that change will simply have to happen, that businesses will deftly re-organize, that health care will remake itself for efficiency and better outcomes, that people will adopt new behaviors that make them happier, healthier, more prosperous and peaceful. Maybe. But almost everything we know about technology and society across human history argues that it won’t be so straightforward.

Data Science is becoming mature enough to grapple confidently and creatively with humans, with organizations, with the power of archaic conventions that societies are stuck following. The field is broadening to a place where data science is becoming as much a social scientific endeavor as a technical one. The next generation of world class data scientists will need the technical skills to work with huge amounts of data, the analytical skills to understand how it is embedded in business and society, and the design and storytelling skills to pull these insights together and use them to motivate change.

What skills, knowledge, and experience do you and your organization need to thrive in a data-intensive economy? Come join senior industry and academic leaders at DataEDGE at UC Berkeley on May 30-31 to engage in what will be a lively and important conversation aimed at answering today’s questions about the data science revolution—and formulating tomorrow’s.


Continua a leggere