In Data Science, making a generic model which can help us understand data or represent data well is our main motto. So, for preparing model we have lots of Algorithms and techniques. But what we lack is knowing which the right one for us.
Of course, with respect to our my limited knowledge of understanding or exploring the world of machine learning algorithms we can’t absolutely/perfectly pinpoint what algorithms that we want but with what we know, I have these design principles to help your choose better.
3 Generic Rules that are applicable in all machine learning models
1. Ockham’s Razor
“Entities should not be multiplied beyond necessity. Plurality should not be posited without necessity.”
As Leonardo da Vinci remarked, “Simplicity is the ultimate sophistication.” In design application, Ockham’s Razor is a warning to avoid over complicating design, information or instructions.
2. Hicks Law
Hick’s Law, the product of British psychologist William Edmund Hick in the 1950’s, indirectly supports Ockham’s Razor. This Law suggests that the more choices presented, the longer and more difficult will be the decision-making process.
“The time it takes to make a decision increases as the number of alternatives increases.”
3. Fitts’ Law
Fitts’ Law, the brainchild of Paul Fitts in 1954, draws a connection between the length of time required to move to a target and the size and distance of that target.
“The easier the target is to identify and the shorter the distance, the quicker the task can be performed, and the more pleasant the experience.”
3 Natures Rules that are seen almost everywhere
4. Fibonacci Sequence
The Fibonacci Sequence, named for Leonardo of Pisa, who also went by the name Fibonacci, was first identified in the early 1200s, but previously understood by older Indian mathematicians. This list of numbers is created by repeatedly adding the sums of the last two digits. It begins with 0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144 and so on. What is fascinating is that this sequence appears repeatedly in nature, in tree branches, leaves on stems, in a wide variety of plants and the reproduction pattern of honeybees.
5. Golden Ratio
The Golden Ratio is actually similar in mathematical relationship to the Fibonacci Sequence. The golden ratio is an irrational mathematical constant, approximately 1.6180339887. Understanding of this principle dates back more than 2000 years, but it was Euclid who first gave it a definition. The Golden Ratio describes the most aesthetically pleasing proportionate shapes and designs.
Basically, “The sum of the quantities of the larger quantity will equal the ratio of the larger quantity to the smaller one.”
Also called the “Golden Rectangle” this ratio describes the relationship between the longer side and the shorter side of a rectangle.
Post cards, playing cards, movie posters and even innocuous light switch plates take advantage of the subconscious approval humans have of this shape. It can be incorporated into product design, advertising engineering, art and many other fields.
The golden ratio, otherwise known as Phi, is considered a ‘comfortable’ ratio, in that it is all around us and the eye has therefore been trained to find it attractive.
6. Rule of Thirds
The Rule of Thirds may at first seem similar to the Golden Ratio, but, in fact, it is not at all the same. This principle is first noted in 1797, in John Thomas Smith’s book Remarks on Rural Scenery. Smith understood his discovery to be a very general principle and not an absolute law. However, the reality is that when design honors this principle, the results are aesthetically attractive to the majority of observers.
Cause & Effect
7. Pareto Principle
The Pareto Principle, also known as the “Law of the Vital Few” or the “80/20 Rule” is well known, even if not by title, to anyone who has the responsibility of managing a workforce, congregation or classroom of students.
The Pareto principle states that, for many events, roughly 80% of the effects come from 20% of the causes.
Logical Thinking or Preparation Rule
8. Mental Model
The Mental Model Law has been around as long as there have been imaginative scientists, but it was Kenneth Craik, in 1943, who proposed that “the human mind constructs small-scale mental models of reality that it uses to anticipate events.” It is certainly true that learning becomes much easier when you can build upon a model of something that is already familiar. Computer designers tapped into this when they incorporated familiar terms such as desk top, files and folders into word processing programs.
Any time you can use a familiar experience as a model or introduction to a new development, the opportunity for success is increased and the stress level is reduced.
Error handling Rules
9. Tolerance for Error
Tolerance for Error, also known as Design for Error is a concept that attempts to minimize the inevitable problems and negative consequences that follow accidental or unintended actions. By incorporating Ockham’s Razor and the rule of simplicity once more, the tendency for errors to happen can be reduced; however, planning for their occasional occurrence is wise on any level.
By arranging information and elements as simply and minimally as possible and by providing warnings and fail safe features you can limit the negative effects of wrong actions.
10. Signal to Noise Ratio
Performance Rule. The Signal to Noise Ratio is actually an identification of the balance between the message and the background noise that could limit it or weaken its impact. When a 1:1 ratio is suggested, this means that the background distractions are competing equally with the intended message or performance. The results will always be less than desired when this happens.
Rule of Communication
11. Equitable and Flexible Use Laws
A promotion or advertisement that appeals to all will definitely produce greater results and be worth the time and effort in design.
Explain things in simpler terms that everyone can understand and appreciate.
12. Law of Perceptible Information
The Law of Perceptible Information practically speaks for itself. Unless information and instructions are communicated in an easy-to-understand manner, effectiveness and positive response will be lost. Many times the best instructions are the shortest and simplest. By using both audio and visual formats, including graphics and verbiage, presentations can be redundant without being boring or insulting.
Using different elements to disseminate important instructions, directions or information is key to laying the ground work for a positive response.
Source: http://www.onextrapixel.com/2011/06/09/12-laws-and-principles-to-aid-you-in-your-design/
Python Data Engineer Notes 