Adonis Diaries

Archive for the ‘Human Factors/Ergonomics’ Category

Just how people behave on Average

Posted on November 21, 2016

Many articles and books have been published for every single one of these principles, effects and laws.

I stumbled on a term-paper that a student of mine submitted in 2002 for the course of Human Factors in Engineering and I said: Why not? It is a good topic to post

Most of these principles were formulated by psychology researchers and they are good guidelines of what to expect in pitfalls and erroneous judgement when designing for people usage.

These laws and principles cannot be classified as rules for solving problems as is commonly misunderstood in natural sciences.

Many of these principles were the results of experiments with failed hypothesis because they were not tightly controlled.

Basically, if you know how average people behave in your community, you can design for effective results

Consequently, the first critical phase in any project is to comprehend the idiosyncrasies of the particular community in order to design valid solutions

First, check the ones you have already heard of, or read about in your course works.

  1. Hawthorn Effect
  2. Placebo Effect
  3. Occam’s razor
  4. Peter principle
  5. Parkinson’s Law
  6. Murphy’s law
  7. Pareto Principle
  8. Rule of Redundant systems
  9. Zeigarnik Effect
  10. Contrast principle
  11. Cognitive Dissonance
  12. Perceptual Consistency
  13. Turnpike Effect

Actually, last year I read a book “How to think clear” and it developed on many of these biases and effects. I reviewed many of the chapters.

Hawthorn Effect

The motivated people have greater effect on the solution presented to resolve a problem.

In the mid 1930’s a vast experiment involved thousands of employees who were supposed to ignore that an experiment is taking place. It turned out that the employees got wind and overdid their best at work. An example of an experiment that was not very well controlled.

Placebo Effect

A harmless with No pharmacological effects may make sick people feeling better if they were told the medicine is part of the cure.

Apparently, placebo has positive effect even though the sick person was told that it is a harmless medicine. (Maybe the sick person doesn’t really believe what he was told?)

William of Occam’s razor

The explanation with the fewest assumptions is the correct alternative in most cases. (Which means that the more you control for many variables, the better you avoid biases and presumptions)

Peter principle

Employee tends to rise to his level of incompetence. When a competent employee rises to a higher level of complexities then they fall back to an incompetent job where they are not positioned to fill.

Parkinson’s law:

Work expands to fill the time allotted to it: The procrastination effect.

Any work must be subdivided to last a definite time span so that the entire project is finished according to a timetable and on schedule.

Give a student a project that can be done within a few days and he will gladly leave it to the last minutes after a few months for the scheduled time for presentation.

Murphy’s law

If anything can go wrong, it will go wrong. We tend not to expect what we think is an unexpected event or behaviour.

Pareto Principle

A small fraction of people do most of the job. The wealthiest are a tiny fraction of the total population. A fraction of the items sold generate most of the profit or revenue.

Rule of Redundant systems

Every critical system requires a redundant backup system. (The danger is that the more you include redundant backups, the more the system become complicated and tends to breakdown more often)

Zeigarnik Effect

We prefer to have a closure on a task before starting another one. Handling simultaneous tasks is difficult for most people and they are upset when they are asked to interrupt a job in midstream in order to tend to another job.

Contrast principle

The last event in a stream of successive events is retained and valued more than any of the other events. If the latest person seemed nice, he is viewed as nicer than he is. A good suggestion offered after a series of bad suggestions feels better than it is.

Cognitive Dissonance

Hearing about a crime committed creates a dissonance in the belief system of morality and justice and the event that occurred.

If we believe that a certain event should not happen then we tend to find fault in the victim.

Perceptual Consistency

We tend to pigeon-hole people and circumstances into simple generalized entities.

Turnpike Effect

The availability of unforeseen utility of a resource or facility generates greater use than was predicted.

Improve the road condition of a side route and people will drive on it more frequently than expected.

Guess what my job is: Human Factors in Engineering?

Posted on June 25, 2009 (Written in November 13, 2005)

“Guess what my job is”

It would be interesting to have a talk with the freshly enrolled engineering students from all fields as to the objectives and meaning of designing products, projects and services.

This talk should be intended to orient engineers for a procedure that might provide their design projects the necessary substance for becoming marketable and effective in reducing the pitfalls in having to redesign for failing to consider the health and safety of what they produced and conceived.

This design behavior should start right at the freshman level while taking formal courses so that prospective engineers will naturally apply this acquired behavior in their engineering career.

In the talk, the students will have to guess what the Human Factors discipline is from the case studies, exercises and problems that will be discussed.

The engineers will try to answer a few of the questions that might be implicit, but never formally explicitly explained or learned in engineering curriculums, because the necessary courses are generally offered outside their traditional discipline field.

A sample of the questions might be as follows:

1. What is the primary job of an engineer?

2. What does design means?  How do you perceive designing to look like?

3. To whom are you designing?  What category of people?

4. Who are your target users? Engineer, consumers, support personnel, operators?

5. What are your primary criteria in designing?  Error free application product?

6. Who commit errors?  Can a machine do errors?

7. How can we categorize errors?  Any exposure to an error taxonomy?

8. Can you foresee errors, near accidents, accidents?  Take a range oven for example, expose the foreseeable errors and accidents in the design and specifically the display and control idiosyncrasy.

9. Who is at fault when an error is committed or an accident occurs?

10. Can we practically account for errors without specific task taxonomy?

11. Do you view yourself as responsible for designing interfaces to your design projects depending on the target users?

12. Would you relinquish your responsibilities for being in the team assigned to design an interface for your design project?

13. What kinds of interfaces are needed for your design to be used efficiently?

14. How engineers solve problems?  Searching for the applicable formulas? Can you figure out the magnitude of the answer?  Have you memorized the allowable range for your answers from the given data and restriction imposed in the problem after solving so many exercises?

15. What are the factors or independent variables that may affect your design project?

16. How can we account for the interactions among the factors?

17. Have you memorize the dimensions of your design problem?

18. Have you been exposed to reading research papers? Can you understand, analyze and interpret the research paper data? Can you have an opinion as to the validity of an experiment?

19. Would you accept the results of any peer-reviewed article as facts that may be readily applied to your design projects? Can you figure out if the paper is Not biased or extending confounding results?

20. Do you expect to be in charge of designing any new product or program or procedures in your career?

21. Do you view most of your job career as a series of supporting responsibilities; like just applying already designed programs and procedures?

22. Are you ready to take elective courses in psychology, sociology, marketing, and business targeted to learn how to design experiments and know more about the capabilities, limitations and behavioral trends of target users?

23. Are you planning to go for graduate studies?  Do you know what elective courses might suit you better in your career?

Let’s experiment

Posted on November 26, 2010

Whether we admit it or not, every person has constructed a mental model of how he views the universe and life.  

For example, was the universe created, is it infinite, is it timeless… 

And what is life, the purpose of life, what happens after death, is there a soul, what happens to the soul, is the soul individual or a collective soul…?

Since antiquity, philosophers have been discussing and reasoning on the following matter:  

“Do mankind enjoys an innate general spirit (regardless of ethnicity, culture, gender…) that expresses how he views the construct of the universe, or it is an individual learning process relevant to the manner the various sensory organs observe nature and people and organize the information”?

The hypothesis is

Do people with sensory handicaps (blind, deaf…) extend the same kind of subjective understanding of the universe and life as “normal” people do, across all ethnic cultures with oral and written myths and traditions?

First, we need baseline stories on “What do I know about the universe and life?” from “normal” people with “normally” functioning sensory organs (vision, audition…). 

The baseline stories should be captured from varieties of ethnic cultural entities in the five continents, privileging the oral cultures with No recognized written documents, and minority cultures with written cultures but Not read or disseminated universally. 

The baseline stories must discriminate between genders (between group factors) and the ethnic stories within each gender groups.

The baseline stories must discriminate among the stage of maturity of the storyteller (young, adult, middle age, and older people). 

The baseline stories must discriminate among the literacy levels of the subjects (such as they read and write in one language, read only, and only orally literate subjects).  Thus, the team of experimenters must be trained to adequately record answers and stories in uniform fashion.

The next phase of the experiment is gathering stories of sensory handicapped people in the above ethnic and gender groups (blind, deaf…)

We may extend this experiment by artificially handicapping a normal subject by preventing him to see or to hear while resuming his “normal” live for a period.  Do you think that his mental model of the universe might be altered significantly?

Another extension may be involving normal sensory subjects but with different mental capabilities and limitations (over developed or under developed brain powers).  

This experiment would answer the question: “Are reading and listening to stories generate different types of observational data due to further brain processing mechanisms?”

The most essential preparation for the experiment is the designing of an exhaustive questionnaire with exhaustive options to educate the subjects on the varieties of viewpoints and myths. 

For that purpose, the questionnaire will be tested on many preliminary samples of ethnic cultures in order to catch and collect the varieties of relevant options, sort of exhaustive compendium on the different myths and mental models. 

I would recommend that the design requires every question to be answered. This means that those logical procedures of demanding the subject to skip several questions, as in filling tax forms, be eliminated:  We should not fall in the bias of enforcing our rational logic on oral culture ethnic groups and the illiterates.

It is advisable that follow-up oral stories accompany answering the questionnaire. Then, another follow-up written story be attached to the oral story. 

The written story would condense the individual story into a comprehensive and coherent story after the preceding two educational sessions. 

The teams of trained experimenters would have to fill the initial questionnaire with the new information generated by the oral and written stories; missing information can be filled by default, using the original questionnaire for each subject. 

Thus, data analysis can be conducted on the two questionnaires: the before learning process and the after learning process of the mental models.

I find it interesting that, after the written story, the subject would give his opinion on the current theories of astrophysicists on the universe in order to check the cohesion and compatibility of the subjects in their perception of the universe. 

For example: what they think of the theory that this universe is the product of a collision between two universes; that a universe revolves around each black hole; that what we see is a simulated universe of a matrix universe; that the sky is a wall on which the image of the stars and galaxies are projected onto it (a universe as hologram); that the universe keeps changing every time we observe it… 

Do you think that you might change your view if a theory (coming from an astrophysicist) impresses you?

The spontaneous suggestion is “why not ask a subject to tell his story before answering a questionnaire? At least we can have an original version, unbiased by constructed questionnaires.”  

This suggestion is pertinent if it is feasible to nudge a subject to start telling a story without a prompt sheet containing the necessary lines of thoughts to guide the subject in the endeavor: The prompt sheet must be devoid of any biased suggestions.  

In any case, I believe that devising such a prompt sheet is necessary, even if not applied in the experiment, in order to get the questionnaire developed and cleaned of idiosyncratic allusions and local imageries.

The experiment is complex and will need to be subdivided in meaningful stages of shorter experiments.

It is time intensive and for a long duration.

It requires training of large teams of researchers and experimenters.  Preliminary experiments would show the best ways of experimenting piece meal this vast project.

Note 1:  I tend to include materials we read and stories we heard as sensory inputs since they are processed by the brain, at various levels, as sensory observations.

Note 2: Many scholars present the view that what we actually sense are in fact “processed observations”, and not the raw sensed data, since all sensing observations are data processed by the brain at different levels of manipulations.

Good enough: We are dealing with what mankind is observing: That is what is available to forming a coherent structure of the universe and the environment we live into. 

The follow-up lesson is:  Other reasoning species must be viewing the universe differently since their senses have different capacities and limitations, and their brain structures are different from mankind.

Note 3:  The essential question that the previous experiment might offer an answer to is:  “If an individual is handicapped in one or more sensory organs then, by reading or listening to stories, can his brain re-establish what normal people comprehend of the universe?”

Note 4: I conjecture that all the facts, observations, experiments., philosophy… will Not tell us anything “sustainable” of what is life and the universe. What this experiment could boils down to is to “know”:

How the majority, in any ethnic group, likes to conceive the nature of Life and the Universe?

This is fundamental to to evaluate the evolution of human “Emotional Intelligence

An exercise: taxonomy of methods

Posted on: June 10, 2009

Article #14 in Human Factors

I am going to let you have a hand at classifying methods by providing a list of various methods that could be used in Industrial engineering, Human Factors, Ergonomics, and Industrial Psychology.

This first list of methods is organized in the sequence used to analyzing part of a system or a mission;

The second list is not necessarily randomized, though thrown in without much order; otherwise it will not be an excellent exercise.

First, let us agree that a method is a procedure or a set of step by step process that our forerunners of geniuses and scholars have tested, found it good, agreed on it on consensus basis and offered it for you to use for the benefit of progress and science.

Many of you will still try hard to find short cuts to anything, including methods, for the petty argument that the best criterion to discriminating among clever people is who waste time on methods and who are nerds.

Actually, the main reason I don’t try to teach many new methods in this course (Human Factors in Engineering) is that students might smack run into a real occupational stress, which they are Not immune of, especially that methods in human factors are complex and time consuming.

Here is this famous list of a few methods and you are to decide which ones are still in the conceptual phases and which have been “operationalized“.

The first list contains the following methods:

Operational analysis, activity analysis, critical incidents, function flow, decision/action, action/information analyses, functional allocation, task, fault tree, failure modes and effects analyses, timeline, link analyses, simulation, controlled experimentation,  operational sequence analysis, and workload assessment.

The second list is constituted of methods that human factors are trained to utilize if need be such as:

Verbal protocol, neural network, utility theory, preference judgments, psycho-physical methods, operational research, prototyping, information theory, cost/benefit methods, various statistical modeling packages, and expert systems.

Just wait, let me resume.

There are those that are intrinsic to artificial intelligence methodology such as:

Fuzzy logic, robotics, discrimination nets, pattern matching, knowledge representation, frames, schemata, semantic network, relational databases, searching methods, zero-sum games theory, logical reasoning methods, probabilistic reasoning, learning methods, natural language understanding, image formation and acquisition, connectedness, cellular logic, problem solving techniques, means-end analysis, geometric reasoning system, algebraic reasoning system.

If your education is multidisciplinary you may catalog the above methods according to specialty disciplines such as:

Artificial intelligence, robotics, econometrics, marketing, human factors, industrial engineering, other engineering majors, psychology or mathematics.

The most logical grouping is along the purpose, input, process/procedure, and output/product of the method. Otherwise, it would be impossible to define and understand any method.

Methods could be used to analyze systems, provide heuristic data about human performance, make predictions, generate subjective data, discover the cause and effects of the main factors, or evaluate the human-machine performance of products or systems.

The inputs could be qualitative or quantitative such as declarative data, categorical, or numerical and generated from structured observations, records, interviews, questionnaires, computer generated or outputs from prior methods.

The outputs could be point data, behavioral trends, graphical in nature, context specific, generic, or reduction in alternatives.

The process could be a creative graphical or pictorial model, logical hierarchy or in network alternative, operational, empirical, informal, or systematic.

You may also group these methods according to their mathematical branches such as algebraic, probabilistic, or geometric.

You may collect them as to their deterministic, statistical sampling methods and probabilistic characters.

You may differentiate the methods as belonging to categorical, ordinal, discrete or continuous measurements.

You may wish to investigate the methods as parametric, non parametric, distribution free population or normally distributed.

You may separate them on their representation forms such as verbal, graphical, pictorial, or in table.

You may discriminate them on heuristic, observational, or experimental scientific values.

You may bundle these methods on qualitative or quantitative values.

You may as well separate them on their historical values or modern techniques based on newer technologies.

You may select them as to their state of the art methods such as ancient methods that new information and new paradigms have refuted their validity or recently developed.

You may define the methods as those digitally or analytically amenable for solving problems.

You may choose to draw several lists of those methods that are economically sounds, esoteric, or just plainly fuzzy sounding.

You may opt to differentiate these methods on requiring high level of mathematical reasoning that are out of your capability and those that can be comprehended through persistent efforts.

You could as well sort them according to which ones fit nicely into the courses that you have already taken, but failed to recollect that they were indeed methods worth acquiring for your career.

You may use any of these taxonomies to answer an optional exam question with no guarantees that you might get a substantial grade.

It would be interesting to collect statistics on how often these methods are being used, by whom, for what rational and by which line of business and by which universities.

It would be interesting to translate these methods into Arabic, Chinese, Japanese, Hindu, or Russian.

Has Big Brother no longer a need to disguise his dominion?

December 15, 2020

Face recognition, surveillance concepts. Hand holding smartphone with watching eye on screen. Mobile phone with eye icon. Modern flat design, vector illustration.

Phone is watching you art concept. “You had to live—did live, from habit that became instinct—in the assumption that every sound you made was overheard, and, except in darkness, every movement scrutinized.”—George Orwell, 1984 […]

It had the potential for disaster.

Early in the morning of Monday, December 15, 2020, Google suffered a major worldwide outage in which all of its internet-connected services crashed, including Nest, Google Calendar, Gmail, Docs, Hangouts, Maps, Meet and YouTube.

The outage only lasted an hour, but it was a chilling reminder of how reliant the world has become on internet-connected technologies to do everything from unlocking doors and turning up the heat to accessing work files, sending emails and making phone calls.

A year earlier, a Google outage resulted in Nest users being unable to access their Nest thermostats, Nest smart locks, and Nest cameras.

As Fast Company reports, “This essentially meant that because of a cloud storage outagepeople were prevented from getting inside their homes, using their AC, and monitoring their babies.”

Welcome to the Matrix.

Twenty-some years after the Wachowskis’ iconic film, The Matrix, introduced us to a futuristic world in which humans exist in a computer-simulated non-reality powered by authoritarian machines—a world where the choice between existing in a denial-ridden virtual dream-state or facing up to the harsh, difficult realities of life comes down to a blue pill or a red pill—we stand at the precipice of a technologically-dominated matrix of our own making.

We are living the prequel to The Matrix with each passing day, falling further under the spell of technologically-driven virtual communities, virtual realities and virtual conveniences managed by artificially intelligent machines that are on a fast track to replacing human beings and eventually dominating every aspect of our lives.

Science fiction has become fact.

In The Matrixcomputer programmer Thomas Anderson a.k.a. hacker Neo is wakened from a virtual slumber by Morpheus, a freedom fighter seeking to liberate humanity from a lifelong hibernation state imposed by hyper-advanced artificial intelligence machines that rely on humans as an organic power source.

With their minds plugged into a perfectly crafted virtual reality, few humans ever realize they are living in an artificial dream world.

Neo is given a choice: to take the red pill, wake up and join the resistance, or take the blue pill, remain asleep and serve as fodder for the powers-that-be.

Most people opt for the blue pill.

In our case, the blue pill—a one-way ticket to a life sentence in an electronic concentration camp—has been honey-coated to hide the bitter aftertaste, sold to us in the name of expediency and delivered by way of blazingly fast Internet, cell phone signals that never drop a call, thermostats that keep us at the perfect temperature without our having to raise a finger, and entertainment that can be simultaneously streamed to our TVs, tablets and cell phones.

Yet we are not merely in thrall with these technologies that were intended to make our lives easier. We have become enslaved by them.

Look around you. Everywhere you turn, people are so addicted to their internet-connected screen devices—smart phones, tablets, computers, televisions—that they can go for hours at a time submerged in a virtual world where human interaction is filtered through the medium of technology.

This is not freedom.

This is not even progress.

Big Brother in Disguise: The Rise of a New, Technological World Order

By Kenneth T.

My blog, My way

Welcome to a little piece of my life.
Here you will find things concerning my everyday experiences and or my thoughts on everyday happenings.

For instance you may find thoughts of my Farmstead, which is as my wife calls it, our Accidental Farming life.

Perhaps on a whim, I might just jump on a soap box about what’s going on with my crazy family (the immediate one, that is).~You don’t need to put a penny in the coin slot for any commentary there~

You may find, new additions to what I call “Hobby-time”. I make pinback buttons (some call them badges).

And then there is the outside the box or “Off-track” thinking, part of me. Which can be anything else from aliens to the zoology of the Loch Ness monster…

This will probably be more mundane as health concerns, for instance, to vaccinate or not.

Is the Earth Flat or is it Hollow? Is there a dome? Is any of it real? Do you really want to know?

Police brutality and the continuing corruption of established government, Big Business, Big Oil, Big Brother. Can we survive?

Should we survive? The coming monetary collapse.
There is so much going on, more than we see outside our windows.View Archive →

A good time to die (October 16, 2008)

We know by now that decisions for resuming experiments on atomic explosions, in open air or underground, are bad news.  

We know that decisions to leave man out of the loop of programmed launching of guided ballistic missiles are wrong decisions.  

We are learning that the ozone layer is good and protects the living organisms from lethal doses of ultraviolet radiations; that the depletion of ozone over the Antarctic is very bad news.  

We recognize that the increased concentration of CO2 may be causing the “Greenhouse Effect”, melting the North Pole and increasing the Oceans water level.  (Methane increased emission from the poles from the melting of permafrost layer is extremely bad news)

We have this gut feeling that the deforestation of the virgin forests in the Equator is degrading the quality of air and increasing the numbers of tsunamis or cyclones or tidal waves or hurricanes.  

We blame those who still insist on residing around the targeted sea shores as if these cataclysms would disappear any time soon.  

We are less sure how the high tension pylons amidst towns alter the health of children. Active citizens must have learned the lesson to no longer wait for the results of funded research by multinationals and experiments when health and safety are of concern.

We know that our intelligence is intrinsically malignant, but the most malignant are those vicious, lengthy and recurring cycles of the decision processes to settle on remedial plans of actions.

We frequently don’t know the mechanisms to resolve what we initiated and much less these processes that takes decades to recognize the problems and reach agreements to act upon and persevere in our programs.  

Earth has mechanisms to stabilize harms done to it, but it requires man to leave it alone for hundreds and thousands of years.

Every time man creates a problem to earth’s quality and stability we have to wait for a valiant scientist to sound the alarm.  

Then we have to wait for this scientist to affiliate with a recognized international figure to give credit and weight for his discovery.  

Then we have to wait for the convinced scientists and professionals to sign up a manifest and present it to the UN so that the UN might receives a wake up call to take on its responsibilities in order to preserve human rights for clean air, clean potable water, clean environment and human rights for health and safety and security.  

Then we have to wait for one superpower to admit that what is happening is bad, that the level of tolerance, invariably set by unprofessional specialists in the field, is no longer acceptable.  

Then we have to wait for one superpower to unilaterally agree to distance itself from the pack of wolves and actively remediate.

Then we have to hear the complaints of economic infeasibility of regulations to remedial actions and

Then we have to set a period that lengthens to decades to start an effective program that agrees to everyone concerned.

Albert Schweitzer in his book of selected three calls to action “Peace or atomic war” describes the fundamental process that was initiated to put a halt on live atomic explosion experimentations.  

You discover that physicists and not medical specialists volunteer to set levels of tolerances to radioactive emissions.  

You hear Edward Teller, the “eminent” physicist and “father” of the hydrogen bomb say “We have got for our national security to keep testing for a harmless hydrogen bomb”; as if States at war intend not to inflict harms!  

The UN had to wait for 9235 scientists and headed by Linus Pauling to sign a manifest in January 1958 explaining the lethal harm to the next generations of radioactive emissions.  

Then the US Administration gradually stopped financing apologetics in Newspapers that the experiments constitute no tangible harms.

De Gaulle of France sank an entire atole in the Pacific to test His open nuclear bomb. The French operators (in shorts and naked chest) and the people in the adjacent islands were Not warned. Most of them died from Not natural causes.

16,000 US navy personnels on a destroyer were ordered to turn their faces into a direction and cover the faces. They were Not warned that a nuclear test is going to be experimented. The marines could see the bones of their comrades from the X-rays and many were blown off. 15,000 of them died, and Not from natural causes.

After the US, Britain and the Soviet Union were forced to agree on a moratorium to open air explosions they resumed their nuclear explosions in “controlled, secure, and safe” underground testing fields

I never stumbled on a manuscript describing the consequences for underground nuclear testing.  

Usually the consequences are of long term nature and time-line researches are too expensive to follow up.  

My gut feeling is that these underground testing are directly linked to the current drastic increase in large scale seism, volcano eruptions and tidal wave catastrophes.  

Earth may sustain one major destructive factor but it requires more than one main factor to destabilize earth and its environment.

Which machine learning algorithm should I use? How many and which one is best?

Note: in the early 1990’s, I took graduate classes in Artificial Intelligence (AI) (The if…Then series of questions and answer of experts in their fields of work) and neural networks developed by psychologists. 

The concepts are the same, though upgraded with new algorithms and automation.

I recall a book with a Table (like the Mendeleev table in chemistry) that contained the terms, mental processes, mathematical concepts behind the ideas that formed the AI trend…

There are several lists of methods, depending on the field of study you are more concerned with.

One list of methods is constituted of methods that human factors are trained to utilize if need be, such as:

Verbal protocol, neural network, utility theory, preference judgments, psycho-physical methods, operational research, prototyping, information theory, cost/benefit methods, various statistical modeling packages, and expert systems.

There are those that are intrinsic to artificial intelligence methodology such as:

Fuzzy logic, robotics, discrimination nets, pattern matching, knowledge representation, frames, schemata, semantic network, relational databases, searching methods, zero-sum games theory, logical reasoning methods, probabilistic reasoning, learning methods, natural language understanding, image formation and acquisition, connectedness, cellular logic, problem solving techniques, means-end analysis, geometric reasoning system, algebraic reasoning system.

Hui Li on Subconscious Musings posted on April 12, 2017 Advanced Analytics | Machine Learning

This resource is designed primarily for beginner to intermediate data scientists or analysts who are interested in identifying and applying machine learning algorithms to address the problems of their interest.

typical question asked by a beginner, when facing a wide variety of machine learning algorithms, is “which algorithm should I use?”

The answer to the question varies depending on many factors, including:

  • The size, quality, and nature of data.
  • The available computational time.
  • The urgency of the task.
  • What you want to do with the data.

Even an experienced data scientist cannot tell which algorithm will perform the best before trying different algorithms.

We are not advocating a one and done approach, but we do hope to provide some guidance on which algorithms to try first depending on some clear factors.

The machine learning algorithm cheat sheet

Flow chart shows which algorithms to use when

The machine learning algorithm cheat sheet helps you to choose from a variety of machine learning algorithms to find the appropriate algorithm for your specific problems.

This article walks you through the process of how to use the sheet.

Since the cheat sheet is designed for beginner data scientists and analysts, we will make some simplified assumptions when talking about the algorithms.

The algorithms recommended here result from compiled feedback and tips from several data scientists and machine learning experts and developers.

There are several issues on which we have not reached an agreement and for these issues we try to highlight the commonality and reconcile the difference.

Additional algorithms will be added in later as our library grows to encompass a more complete set of available methods.

How to use the cheat sheet

Read the path and algorithm labels on the chart as “If <path label> then use <algorithm>.” For example:

  • If you want to perform dimension reduction then use principal component analysis.
  • If you need a numeric prediction quickly, use decision trees or logistic regression.
  • If you need a hierarchical result, use hierarchical clustering.

Sometimes more than one branch will apply, and other times none of them will be a perfect match.

It’s important to remember these paths are intended to be rule-of-thumb recommendations, so some of the recommendations are not exact.

Several data scientists I talked with said that the only sure way to find the very best algorithm is to try all of them.

(Is that a process to find an algorithm that matches your world view on an issue? Or an answer that satisfies your boss?)

Types of machine learning algorithms

This section provides an overview of the most popular types of machine learning. If you’re familiar with these categories and want to move on to discussing specific algorithms, you can skip this section and go to “When to use specific algorithms” below.

Supervised learning

Supervised learning algorithms make predictions based on a set of examples.

For example, historical sales can be used to estimate the future prices. With supervised learning, you have an input variable that consists of labeled training data and a desired output variable.

You use an algorithm to analyze the training data to learn the function that maps the input to the output. This inferred function maps new, unknown examples by generalizing from the training data to anticipate results in unseen situations.

  • Classification: When the data are being used to predict a categorical variable, supervised learning is also called classification. This is the case when assigning a label or indicator, either dog or cat to an image. When there are only two labels, this is called binary classification. When there are more than two categories, the problems are called multi-class classification.
  • Regression: When predicting continuous values, the problems become a regression problem.
  • Forecasting: This is the process of making predictions about the future based on the past and present data. It is most commonly used to analyze trends. A common example might be estimation of the next year sales based on the sales of the current year and previous years.

Semi-supervised learning

The challenge with supervised learning is that labeling data can be expensive and time consuming. If labels are limited, you can use unlabeled examples to enhance supervised learning. Because the machine is not fully supervised in this case, we say the machine is semi-supervised. With semi-supervised learning, you use unlabeled examples with a small amount of labeled data to improve the learning accuracy.

Unsupervised learning

When performing unsupervised learning, the machine is presented with totally unlabeled data. It is asked to discover the intrinsic patterns that underlies the data, such as a clustering structure, a low-dimensional manifold, or a sparse tree and graph.

  • Clustering: Grouping a set of data examples so that examples in one group (or one cluster) are more similar (according to some criteria) than those in other groups. This is often used to segment the whole dataset into several groups. Analysis can be performed in each group to help users to find intrinsic patterns.
  • Dimension reduction: Reducing the number of variables under consideration. In many applications, the raw data have very high dimensional features and some features are redundant or irrelevant to the task. Reducing the dimensionality helps to find the true, latent relationship.

Reinforcement learning

Reinforcement learning analyzes and optimizes the behavior of an agent based on the feedback from the environment.  Machines try different scenarios to discover which actions yield the greatest reward, rather than being told which actions to take. Trial-and-error and delayed reward distinguishes reinforcement learning from other techniques.

Considerations when choosing an algorithm

When choosing an algorithm, always take these aspects into account: accuracy, training time and ease of use. Many users put the accuracy first, while beginners tend to focus on algorithms they know best.

When presented with a dataset, the first thing to consider is how to obtain results, no matter what those results might look like. Beginners tend to choose algorithms that are easy to implement and can obtain results quickly. This works fine, as long as it is just the first step in the process. Once you obtain some results and become familiar with the data, you may spend more time using more sophisticated algorithms to strengthen your understanding of the data, hence further improving the results.

Even in this stage, the best algorithms might not be the methods that have achieved the highest reported accuracy, as an algorithm usually requires careful tuning and extensive training to obtain its best achievable performance.

When to use specific algorithms

Looking more closely at individual algorithms can help you understand what they provide and how they are used. These descriptions provide more details and give additional tips for when to use specific algorithms, in alignment with the cheat sheet.

Linear regression and Logistic regression

Linear regressionLogistic regression

Linear regression is an approach for modeling the relationship between a continuous dependent variable [Math Processing Error]y and one or more predictors [Math Processing Error]X. The relationship between [Math Processing Error]y and [Math Processing Error]X can be linearly modeled as [Math Processing Error]y=βTX+ϵ Given the training examples [Math Processing Error]{xi,yi}i=1N, the parameter vector [Math Processing Error]β can be learnt.

If the dependent variable is not continuous but categorical, linear regression can be transformed to logistic regression using a logit link function. Logistic regression is a simple, fast yet powerful classification algorithm.

Here we discuss the binary case where the dependent variable [Math Processing Error]y only takes binary values [Math Processing Error]{yi∈(−1,1)}i=1N (it which can be easily extended to multi-class classification problems).

In logistic regression we use a different hypothesis class to try to predict the probability that a given example belongs to the “1” class versus the probability that it belongs to the “-1” class. Specifically, we will try to learn a function of the form:[Math Processing Error]p(yi=1|xi)=σ(βTxi) and [Math Processing Error]p(yi=−1|xi)=1−σ(βTxi).

Here [Math Processing Error]σ(x)=11+exp(−x) is a sigmoid function. Given the training examples[Math Processing Error]{xi,yi}i=1N, the parameter vector [Math Processing Error]β can be learnt by maximizing the Pyongyang said it could call off the talks, slated for June 12, if the US continues to insist that it give up its nuclear weapons. North Korea called the military drills between South Korea and the US a “provocation,” and canceled a meeting planned for today with South Korea.of [Math Processing Error]β given the data set.Group By Linear RegressionLogistic Regression in SAS Visual Analytics

Linear SVM and kernel SVM

Kernel tricks are used to map a non-linearly separable functions into a higher dimension linearly separable function. A support vector machine (SVM) training algorithm finds the classifier represented by the normal vector [Math Processing Error]w and bias [Math Processing Error]b of the hyperplane. This hyperplane (boundary) separates different classes by as wide a margin as possible. The problem can be converted into a constrained optimization problem:
[Math Processing Error]minimizew||w||subject toyi(wTXi−b)≥1,i=1,…,n.

A support vector machine (SVM) training algorithm finds the classifier represented by the normal vector  and bias  of the hyperplane. This hyperplane (boundary) separates different classes by as wide a margin as possible. The problem can be converted into a constrained optimization problem:

Linear and kernel SVM charts

When the classes are not linearly separable, a kernel trick can be used to map a non-linearly separable space into a higher dimension linearly separable space.

When most dependent variables are numeric, logistic regression and SVM should be the first try for classification. These models are easy to implement, their parameters easy to tune, and the performances are also pretty good. So these models are appropriate for beginners.

Trees and ensemble trees

A decision tree for prediction model.

Decision trees, random forest and gradient boosting are all algorithms based on decision trees.

There are many variants of decision trees, but they all do the same thing – subdivide the feature space into regions with mostly the same label. Decision trees are easy to understand and implement.

However, they tend to over fit data when we exhaust the branches and go very deep with the trees. Random Forrest and gradient boosting are two popular ways to use tree algorithms to achieve good accuracy as well as overcoming the over-fitting problem.

Neural networks and deep learning

Neural networks flourished in the mid-1980s due to their parallel and distributed processing ability.

Research in this field was impeded by the ineffectiveness of the back-propagation training algorithm that is widely used to optimize the parameters of neural networks. Support vector machines (SVM) and other simpler models, which can be easily trained by solving convex optimization problems, gradually replaced neural networks in machine learning.

In recent years, new and improved training techniques such as unsupervised pre-training and layer-wise greedy training have led to a resurgence of interest in neural networks.

Increasingly powerful computational capabilities, such as graphical processing unit (GPU) and massively parallel processing (MPP), have also spurred the revived adoption of neural networks. The resurgent research in neural networks has given rise to the invention of models with thousands of layers.

A neural network

Shallow neural networks have evolved into deep learning neural networks.

Deep neural networks have been very successful for supervised learning.  When used for speech and image recognition, deep learning performs as well as, or even better than, humans.

Applied to unsupervised learning tasks, such as feature extraction, deep learning also extracts features from raw images or speech with much less human intervention.

A neural network consists of three parts: input layer, hidden layers and output layer. 

The training samples define the input and output layers. When the output layer is a categorical variable, then the neural network is a way to address classification problems. When the output layer is a continuous variable, then the network can be used to do regression.

When the output layer is the same as the input layer, the network can be used to extract intrinsic features.

The number of hidden layers defines the model complexity and modeling capacity.

Deep Learning: What it is and why it matters

k-means/k-modes, GMM (Gaussian mixture model) clustering

K Means ClusteringGaussian Mixture Model

Kmeans/k-modes, GMM clustering aims to partition n observations into k clusters. K-means define hard assignment: the samples are to be and only to be associated to one cluster. GMM, however define a soft assignment for each sample. Each sample has a probability to be associated with each cluster. Both algorithms are simple and fast enough for clustering when the number of clusters k is given.

DBSCAN

A DBSCAN illustration

When the number of clusters k is not given, DBSCAN (density-based spatial clustering) can be used by connecting samples through density diffusion.

Hierarchical clustering

Hierarchical partitions can be visualized using a tree structure (a dendrogram). It does not need the number of clusters as an input and the partitions can be viewed at different levels of granularities (i.e., can refine/coarsen clusters) using different K.

PCA, SVD and LDA

We generally do not want to feed a large number of features directly into a machine learning algorithm since some features may be irrelevant or the “intrinsic” dimensionality may be smaller than the number of features. Principal component analysis (PCA), singular value decomposition (SVD), andlatent Dirichlet allocation (LDA) all can be used to perform dimension reduction.

PCA is an unsupervised clustering method which maps the original data space into a lower dimensional space while preserving as much information as possible. The PCA basically finds a subspace that most preserves the data variance, with the subspace defined by the dominant eigenvectors of the data’s covariance matrix.

The SVD is related to PCA in the sense that SVD of the centered data matrix (features versus samples) provides the dominant left singular vectors that define the same subspace as found by PCA. However, SVD is a more versatile technique as it can also do things that PCA may not do.

For example, the SVD of a user-versus-movie matrix is able to extract the user profiles and movie profiles which can be used in a recommendation system. In addition, SVD is also widely used as a topic modeling tool, known as latent semantic analysis, in natural language processing (NLP).

A related technique in NLP is latent Dirichlet allocation (LDA). LDA is probabilistic topic model and it decomposes documents into topics in a similar way as a Gaussian mixture model (GMM) decomposes continuous data into Gaussian densities. Differently from the GMM, an LDA models discrete data (words in documents) and it constrains that the topics are a priori distributed according to a Dirichlet distribution.

Conclusions

This is the work flow which is easy to follow. The takeaway messages when trying to solve a new problem are:

  • Define the problem. What problems do you want to solve?
  • Start simple. Be familiar with the data and the baseline results.
  • Then try something more complicated.
  • Dr. Hui Li is a Principal Staff Scientist of Data Science Technologies at SAS. Her current work focuses on Deep Learning, Cognitive Computing and SAS recommendation systems in SAS Viya. She received her PhD degree and Master’s degree in Electrical and Computer Engineering from Duke University.
  • Before joining SAS, she worked at Duke University as a research scientist and at Signal Innovation Group, Inc. as a research engineer. Her research interests include machine learning for big, heterogeneous data, collaborative filtering recommendations, Bayesian statistical modeling and reinforcement learning.

Is it the less information the better in critical split-second decision cases?

ER of Cook County Hospital (Chicago) on West Harriston Street, close to downtown, was built at the turn of last century.

I was home of the world’s first blood bank, cobalt-beam therapy, surgeons attaching severed fingers, famous trauma center for gangs’ gunshot wounds and injuries…and most famous for the TV series ER, and George Clooney

In the mid 90’s. the ER welcomed 250,000 patients a year, mostly homeless and health non-insured patients…

Smart patients would come the first thing in the morning to the ER and pack a lunch and a dinner.  Long lines crowded the walls of the cavernous corridors…

There were no air-conditioners: During the summer heat waves, the heat index inside the hospital reached 120 degrees. 

An administrator didn’t last 8 seconds in the middle of one of the wards.

There were no private rooms and patients were separated by plywood dividers.

There were no cafeteria or private phones: The single public phone was at the end of the hall.

One bathroom served all that crowd of patients.

There was a single light switch: You wanted to light a room and the entire hospital had to light up…

The big air fans, the radios and TV that patients brought with them (to keep company), the nurses’ bell buzzing non-stop and no free nurses around… rendered the ER a crazy place to treat emergency cases

Asthma cases were numerous: Chicago was the world worst in patients suffering from asthma…

Protocols had to be created to efficiently treat asthma cases, chest pain cases, homeless patients…

About 30 patients a day converged to the ER complaining of chest pains (potential heart attack worries) and there were only 20 beds in two wards for these cases.

It cost $2,000 a night per bed for serious intensive care, and about $1,000 for the lesser care (nurses instead of cardiologists tending to the chest pain patient…)

A third ward was created as observation unit for half a day patients.  

Was there any rational protocol to decide in which ward the chest-pain patient should be allocated to?

It was the attending physician call, and most of the decisions were wrong, except for the most obvious heart attack cases…

In the 70’s, cardiologist Lee Goldman borrowed the statistical rules of a group of mathematicians for telling apart subatomic particles. Goldman fed a computer data of hundreds of files of heart attack cases and crunched the numbers into a “predictive equation” or model.

Four key risk factors emerged as the most critical telltale of a real heart attack case:

1. ECG (the ancient electrocardiogram graph) showing acute ischemia

2. unstable angina pain

3, fluid in the lungs

4. systolic blood pressure under 100…

decision tree was fine-tuned to decide on serious cases. For example:

1. ECG is normal but at least two key risk factors are positive

2. ECG is abnormal with at leat one risk factor positive…

These kinds of decision trees… (The early artificial programs)

The trouble was that physicians insisted on letting discriminating factors muddle their decisions. For example, statistics had shown that “normally” females do not suffer heart attack until old age, and thus a young female might be sent home (and die the same night) more often than middle-aged black or older white males patients…

Brendan Reilly, chairman of the hospital department of Medicine, decided to try Goldman decision tree.  Physicians were to try the tree and their own instincts for a period.  The results were overwhelmingly in favor of the Goldman algorithm…

It turned out that, if the physician was not bombarded with dozens of pieces of intelligence and just followed the decision tree, he was better off in the allocation to ward process…

For example, a nurse should record all the necessary information of the patients (smoker, age, gender, overweight, job stress, physical activities, high blood pressure, blood sugar content, family history for heart attacks, sweating tendencies, prior heart surgeries,…), but the attending physician must receive quickly the results of the 4 key risk factors to decide on…

Basically, the physician could allocate the patient to the proper ward without even seeing the individual and be influenced by extraneous pieces of intelligence that are not serious today, but could be potential hazards later on or even tomorrow…

Mind you that in order to save on medical malpractice suits, physicians and nurses treating a patient must Not send the patient any signals that can be captured as “contempt”, like feeling invisible and insignificant  https://adonis49.wordpress.com/2012/07/26/what-type-of-hated-surgeons-gets-harassed-with-legal-malpractice-suits/

Many factors are potential predictors for heart attack cases, but they are minor today, for quick decisions…

No need to overwhelm with irrelevant information at critical time.  Analytic reasoning and snap judgment are neither good or bad: Either method is bad at the inappropriate circumstances.

In the “battle field” the less the information coming in, the less the communication streams and the better the rapid cognition decisions of field commanders…

All you need to know is the “forecast” and not the numbers of temperature, wind speed, barometric pressure…

Note: post inspired from a chapter in “Blink” by Malcolm Gladwell

Some have it very easy in life: They are mostly attractive

This Halo effect

A century ago, Edward Lee Thorndike realized that “A single quality or characteristic (beauty, social stature, height…) produces a positive or negative impression that outshine everything else, and the overall effect is disproportionate”

Attractive people have it relatively easy in their professional life and even get better grades from teachers who are affected by the hallo.

Attractive people gets more frequent second chance in life and are believed more frequently than ordinary people.

They get away with many “disappointing” behaviors and performances.

One need not be racist, sexist, chauvinist… to feel victim of this subconscious unjust stereotype.

Otherwise, how can teenagers fall in love and marry quickly?

I have watched many documentaries on the matting processes among animals.

And it was not automatic that the male who danced better, had a louder booming voice, nicer feathers… that won over the females.

Apparently, female animals have additional finer senses to select the appropriate mate.

Have you ever wondered why CEO’s are mostly attractive, tall, with a full chuck of hairs?

Probably because the less attractive are not deemed appropriate for the media?


adonis49

adonis49

adonis49

March 2021
M T W T F S S
1234567
891011121314
15161718192021
22232425262728
293031  

Blog Stats

  • 1,462,324 hits

Enter your email address to subscribe to this blog and receive notifications of new posts by email.adonisbouh@gmail.com

Join 803 other followers

%d bloggers like this: