This document discusses principal component analysis (PCA) and its implementation on Fisher's iris data set. PCA transforms the data to a new coordinate system where the greatest variance by each component is achieved. It was found that the first two principal components explained roughly 96% of the variance in the iris data. PCA was performed by calculating the covariance matrix and performing eigendecomposition to obtain the principal components in order of greatest variance. Projecting the iris data onto the new feature space based on the first two components revealed clusters corresponding to the three iris species.
A Mixed Discrete-Continuous Attribute List Representation for Large Scale Cla...jaumebp
This work assesses the performance of the BioHEL data mining method to handle large-scale datasets, and proposes a representation to deal efficiently with domains with mixed discrete-continuous attributes
In this paper, we are going to introduce the Square root sorting algorithm. We study the best case and worst case of Square
root sorting algorithm, and we compare this algorithm with some of the algorithms that are already existed.
A Sorting Algorithm is used to rearrange a given array elements according to a comparison operator on the elements. So far th ere
are many algorithms that have been used for sorting like: Bubble sort, insertion sort, selection sort, quick sort, merge sort, heap sort
etc. Each of these algorithms are used according to the list of elements of specific usage and they have specific space compl exity
and time complexity as well. The Square root sorting algorithm has the least time complexity comparing to some of the existing
algorithms, especially in case of the best case, and in the worst case it also has less time complexity than some of the existing algorithms, which we will discuss it in coming pages of this paper.
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
ADAPTIVE CONTROL AND SYNCHRONIZATION OF RÖSSLER PROTOTYPE-4 SYSTEMijait
This paper investigates the adaptive control and synchronization of Rössler prototype-4 system with unknown parameters. The Rössler prototype-4 system is a classical three-dimensional chaotic system studied by O.E. Rössler (1979). First, adaptive control laws are designed to stabilize the Rössler prototype-4 system to its unstable equilibrium at the origin based on the adaptive control theory and Lyapunov stability theory. Then adaptive control laws are derived to achieve global chaos synchronization of identical Rössler prototype-4 systems with unknown parameters. Numerical simulations are shown to
validate and illustrate the effectiveness of the proposed adaptive control and synchronization schemes for the Rössler prototype-4 system.
Sorting algorithms are the main concepts of the subject Data Structures and It’s Applications. These
algorithms are designed in arranging the data elements in the sorted order. If the data elements are arranged
in sorted order , then the searching is very easier. Some algorithms are comparison sort and some are noncomparison
sort. The choice of a algorithm is based on the efficiency of the algorithm. I have designed one
algorithm called as Alternate Sort. The main aspect is that different technique of comparisons is involed. I have
presented the algorithm , It’s working and the examples and finally my paper is consisting of the program
listing.
A Mixed Discrete-Continuous Attribute List Representation for Large Scale Cla...jaumebp
This work assesses the performance of the BioHEL data mining method to handle large-scale datasets, and proposes a representation to deal efficiently with domains with mixed discrete-continuous attributes
In this paper, we are going to introduce the Square root sorting algorithm. We study the best case and worst case of Square
root sorting algorithm, and we compare this algorithm with some of the algorithms that are already existed.
A Sorting Algorithm is used to rearrange a given array elements according to a comparison operator on the elements. So far th ere
are many algorithms that have been used for sorting like: Bubble sort, insertion sort, selection sort, quick sort, merge sort, heap sort
etc. Each of these algorithms are used according to the list of elements of specific usage and they have specific space compl exity
and time complexity as well. The Square root sorting algorithm has the least time complexity comparing to some of the existing
algorithms, especially in case of the best case, and in the worst case it also has less time complexity than some of the existing algorithms, which we will discuss it in coming pages of this paper.
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
ADAPTIVE CONTROL AND SYNCHRONIZATION OF RÖSSLER PROTOTYPE-4 SYSTEMijait
This paper investigates the adaptive control and synchronization of Rössler prototype-4 system with unknown parameters. The Rössler prototype-4 system is a classical three-dimensional chaotic system studied by O.E. Rössler (1979). First, adaptive control laws are designed to stabilize the Rössler prototype-4 system to its unstable equilibrium at the origin based on the adaptive control theory and Lyapunov stability theory. Then adaptive control laws are derived to achieve global chaos synchronization of identical Rössler prototype-4 systems with unknown parameters. Numerical simulations are shown to
validate and illustrate the effectiveness of the proposed adaptive control and synchronization schemes for the Rössler prototype-4 system.
Sorting algorithms are the main concepts of the subject Data Structures and It’s Applications. These
algorithms are designed in arranging the data elements in the sorted order. If the data elements are arranged
in sorted order , then the searching is very easier. Some algorithms are comparison sort and some are noncomparison
sort. The choice of a algorithm is based on the efficiency of the algorithm. I have designed one
algorithm called as Alternate Sort. The main aspect is that different technique of comparisons is involed. I have
presented the algorithm , It’s working and the examples and finally my paper is consisting of the program
listing.
Reconstruction of a Complete Dataset from an Incomplete Dataset by ARA (Attri...Waqas Tariq
Preprocessing is crucial steps used for variety of data warehousing and mining Real world data is noisy and can often suffer from corruptions or incomplete values that may impact the models created from the data. Accuracy of any mining algorithm greatly depends on the input data sets. Incomplete data sets have become almost ubiquitous in a wide variety of application domains. Common examples can be found in climate and image data sets, sensor data sets and medical data sets. The incompleteness in these data sets may arise from a number of factors: in some cases it may simply be a reflection of certain measurements not being available at the time; in others the information may be lost due to partial system failure; or it may simply be a result of users being unwilling to specify attributes due to privacy concerns. When a significant fraction of the entries are missing in all of the attributes, it becomes very difficult to perform any kind of reasonable extrapolation on the original data. For such cases, we introduce the novel idea of attribute weightage, in which we give weight to every attribute for prediction of the complete data set from incomplete data sets, on which the data mining algorithms can be directly applied. The attraction behind the idea of weights on attribute and finally averaging it. We demonstrate the effectiveness of the approach on a variety of real data sets. This paper describes a theory and implementation of a new filter ARA (Attribute Relation Analysis) to the WEKA workbench, for finding the complete dataset from an incomplete dataset.
International Journal of Engineering Research and Applications (IJERA) is a team of researchers not publication services or private publications running the journals for monetary benefits, we are association of scientists and academia who focus only on supporting authors who want to publish their work. The articles published in our journal can be accessed online, all the articles will be archived for real time access.
Our journal system primarily aims to bring out the research talent and the works done by sciaentists, academia, engineers, practitioners, scholars, post graduate students of engineering and science. This journal aims to cover the scientific research in a broader sense and not publishing a niche area of research facilitating researchers from various verticals to publish their papers. It is also aimed to provide a platform for the researchers to publish in a shorter of time, enabling them to continue further All articles published are freely available to scientific researchers in the Government agencies,educators and the general public. We are taking serious efforts to promote our journal across the globe in various ways, we are sure that our journal will act as a scientific platform for all researchers to publish their works online.
GLOBAL CHAOS SYNCHRONIZATION OF UNCERTAIN LORENZ-STENFLO AND QI 4-D CHAOTIC S...ijistjournal
In this paper, we apply adaptive control method to derive new results for the global chaos synchronization of 4-D chaotic systems, viz. identical Lorenz-Stenflo(LS) systems (Stenflo, 2001), identical Qi systems (Qi, Chen and Du, 2005) and non-identical LS and Qi systems. In this paper, we shall assume that the parameters of both master and slave systems are unknown and we devise adaptive control schemes for synchronization using the estimates of parameters for both master and slave systems. Our adaptive synchronization schemes derived in this paper are established using Lyapunov stability theory. Since the Lyapunov exponents are not required for these calculations, the adaptive control method is very effective and convenient to synchronize identical and non-identical LS and Qi systems.
Numerical simulations are shown to demonstrate the effectiveness of the proposed adaptive synchronization schemes for the identical and non-identical, uncertain LS and Qi 4-D chaotic systems.
Selection Sort with Improved Asymptotic Time Boundstheijes
Sorting and searching are the most fundamental problems in computer science. Sorting is used for most of the times to help in searching. One of the most well known sorting algorithms that are taught at introductory computer science courses is the classical selection sort. While such an algorithms is easy to explain and grasp at the introductory computer science level, it is far from being an efficient sorting technique, since it requires 푶(풏 ퟐ ) time to sort a list of n numbers. It does so by repeatedly finding the minimum. In this paper we explore the benefit of reducing the search time for the minimum on each pass of the algorithm, and show that we can obtain a worst case time bound of 푶(풏 풏 ퟐ ) by making only minor modifications to the input list. Thus our bound is a factor 푶( 풏 ퟐ ) of faster than the classical selections sort and other classical sorts such as insertion and bubble sort.
Array is a container which can hold a fix number of items and these items should be of the same type. Most of the data structures make use of arrays to implement their algorithms. Following are the important terms to understand the concept of array.
In silico prediction of small molecule properties is widely used today in industry and academia. NMR spectra, in particular, are predicted by a variety of software packages. In this array of software options, two main approaches are used:
Database-based. Compounds are compared against a database, the result is calculated using data for close structural relatives found in the dataset.
Regression-based. An experimental database is used to calculate the parameters for non-linear regression. The chemical shift is calculated by a non-linear function of variables which describe characteristic features of the molecule of interest.
These two outlined approaches require different strategies for implementation and optimization. Database-based results are improved by acquiring larger databases and/or including user-specific data into the calculation. Non-linear regression algorithms can be improved through the regression itself, or by improving the structural descriptors
One of the fundamental issues in computer science is ordering a list of items. Although there is a number of sorting algorithms, sorting problem has attracted a great deal of research, because efficient sorting is important to optimize the use of other algorithms. This paper presents a new sorting algorithm which sort the elements based on their average, which runs faster.. This algorithm was analyzed, implemented and tested and the results are promising for a random data
In this paper a new evolutionary algorithm, for continuous nonlinear optimization problems, is surveyed.
This method is inspired by the life of a bird, called Cuckoo.
The Cuckoo Optimization Algorithm (COA) is evaluated by using the Rastrigin function. The problem is a
non-linear continuous function which is used for evaluating optimization algorithms. The efficiency of the
COA has been studied by obtaining optimal solution of various dimensions Rastrigin function in this paper.
The mentioned function also was solved by FA and ABC algorithms. Comparing the results shows the COA
has better performance than other algorithms.
Application of algorithm to test function has proven its capability to deal with difficult optimization
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
Reconstruction of a Complete Dataset from an Incomplete Dataset by ARA (Attri...Waqas Tariq
Preprocessing is crucial steps used for variety of data warehousing and mining Real world data is noisy and can often suffer from corruptions or incomplete values that may impact the models created from the data. Accuracy of any mining algorithm greatly depends on the input data sets. Incomplete data sets have become almost ubiquitous in a wide variety of application domains. Common examples can be found in climate and image data sets, sensor data sets and medical data sets. The incompleteness in these data sets may arise from a number of factors: in some cases it may simply be a reflection of certain measurements not being available at the time; in others the information may be lost due to partial system failure; or it may simply be a result of users being unwilling to specify attributes due to privacy concerns. When a significant fraction of the entries are missing in all of the attributes, it becomes very difficult to perform any kind of reasonable extrapolation on the original data. For such cases, we introduce the novel idea of attribute weightage, in which we give weight to every attribute for prediction of the complete data set from incomplete data sets, on which the data mining algorithms can be directly applied. The attraction behind the idea of weights on attribute and finally averaging it. We demonstrate the effectiveness of the approach on a variety of real data sets. This paper describes a theory and implementation of a new filter ARA (Attribute Relation Analysis) to the WEKA workbench, for finding the complete dataset from an incomplete dataset.
International Journal of Engineering Research and Applications (IJERA) is a team of researchers not publication services or private publications running the journals for monetary benefits, we are association of scientists and academia who focus only on supporting authors who want to publish their work. The articles published in our journal can be accessed online, all the articles will be archived for real time access.
Our journal system primarily aims to bring out the research talent and the works done by sciaentists, academia, engineers, practitioners, scholars, post graduate students of engineering and science. This journal aims to cover the scientific research in a broader sense and not publishing a niche area of research facilitating researchers from various verticals to publish their papers. It is also aimed to provide a platform for the researchers to publish in a shorter of time, enabling them to continue further All articles published are freely available to scientific researchers in the Government agencies,educators and the general public. We are taking serious efforts to promote our journal across the globe in various ways, we are sure that our journal will act as a scientific platform for all researchers to publish their works online.
GLOBAL CHAOS SYNCHRONIZATION OF UNCERTAIN LORENZ-STENFLO AND QI 4-D CHAOTIC S...ijistjournal
In this paper, we apply adaptive control method to derive new results for the global chaos synchronization of 4-D chaotic systems, viz. identical Lorenz-Stenflo(LS) systems (Stenflo, 2001), identical Qi systems (Qi, Chen and Du, 2005) and non-identical LS and Qi systems. In this paper, we shall assume that the parameters of both master and slave systems are unknown and we devise adaptive control schemes for synchronization using the estimates of parameters for both master and slave systems. Our adaptive synchronization schemes derived in this paper are established using Lyapunov stability theory. Since the Lyapunov exponents are not required for these calculations, the adaptive control method is very effective and convenient to synchronize identical and non-identical LS and Qi systems.
Numerical simulations are shown to demonstrate the effectiveness of the proposed adaptive synchronization schemes for the identical and non-identical, uncertain LS and Qi 4-D chaotic systems.
Selection Sort with Improved Asymptotic Time Boundstheijes
Sorting and searching are the most fundamental problems in computer science. Sorting is used for most of the times to help in searching. One of the most well known sorting algorithms that are taught at introductory computer science courses is the classical selection sort. While such an algorithms is easy to explain and grasp at the introductory computer science level, it is far from being an efficient sorting technique, since it requires 푶(풏 ퟐ ) time to sort a list of n numbers. It does so by repeatedly finding the minimum. In this paper we explore the benefit of reducing the search time for the minimum on each pass of the algorithm, and show that we can obtain a worst case time bound of 푶(풏 풏 ퟐ ) by making only minor modifications to the input list. Thus our bound is a factor 푶( 풏 ퟐ ) of faster than the classical selections sort and other classical sorts such as insertion and bubble sort.
Array is a container which can hold a fix number of items and these items should be of the same type. Most of the data structures make use of arrays to implement their algorithms. Following are the important terms to understand the concept of array.
In silico prediction of small molecule properties is widely used today in industry and academia. NMR spectra, in particular, are predicted by a variety of software packages. In this array of software options, two main approaches are used:
Database-based. Compounds are compared against a database, the result is calculated using data for close structural relatives found in the dataset.
Regression-based. An experimental database is used to calculate the parameters for non-linear regression. The chemical shift is calculated by a non-linear function of variables which describe characteristic features of the molecule of interest.
These two outlined approaches require different strategies for implementation and optimization. Database-based results are improved by acquiring larger databases and/or including user-specific data into the calculation. Non-linear regression algorithms can be improved through the regression itself, or by improving the structural descriptors
One of the fundamental issues in computer science is ordering a list of items. Although there is a number of sorting algorithms, sorting problem has attracted a great deal of research, because efficient sorting is important to optimize the use of other algorithms. This paper presents a new sorting algorithm which sort the elements based on their average, which runs faster.. This algorithm was analyzed, implemented and tested and the results are promising for a random data
In this paper a new evolutionary algorithm, for continuous nonlinear optimization problems, is surveyed.
This method is inspired by the life of a bird, called Cuckoo.
The Cuckoo Optimization Algorithm (COA) is evaluated by using the Rastrigin function. The problem is a
non-linear continuous function which is used for evaluating optimization algorithms. The efficiency of the
COA has been studied by obtaining optimal solution of various dimensions Rastrigin function in this paper.
The mentioned function also was solved by FA and ABC algorithms. Comparing the results shows the COA
has better performance than other algorithms.
Application of algorithm to test function has proven its capability to deal with difficult optimization
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
Have you ever been curious about using Machine Learning and statistical techniques in quantitative finance? This paper gives a brief overview step by step of variable selection, model evaluation, and other insights for when working with financial data
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.
IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit www.ijera.com
Data Trend Analysis by Assigning Polynomial Function For Given Data SetIJCERT
This paper aims at explaining the method of creating a polynomial equation out of the given data set which can be used as a representation of the data itself and can be used to run aggregation against itself to find the results. This approach uses least-squares technique to construct a model of data and fit to a polynomial. Differential calculus technique is used on this equation to generate the aggregated results that represents the original data set.
GENETIC ALGORITHM FOR FUNCTION APPROXIMATION: AN EXPERIMENTAL INVESTIGATIONijaia
Function Approximation is a popular engineering problems used in system identification or Equation
optimization. Due to the complex search space it requires, AI techniques has been used extensively to spot
the best curves that match the real behavior of the system. Genetic algorithm is known for their fast
convergence and their ability to find an optimal structure of the solution. We propose using a genetic
algorithm as a function approximator. Our attempt will focus on using the polynomial form of the
approximation. After implementing the algorithm, we are going to report our results and compare it with
the real function output.
Evaluation of a hybrid method for constructing multiple SVM kernelsinfopapers
Dana Simian, Florin Stoica, Evaluation of a hybrid method for constructing multiple SVM kernels, Recent Advances in Computers, Proceedings of the 13th WSEAS International Conference on Computers, Recent Advances in Computer Engineering Series, WSEAS Press, Rodos, Greece, July 23-25, 2009, ISSN: 1790-5109, ISBN: 978-960-474-099-4, pp. 619-623
A Study on Performance Analysis of Different Prediction Techniques in Predict...IJRES Journal
Time series data is a series of statistical data that is related to a specific instant or a specific time period. Here, the measurements are recorded on a regular basis such as monthly, quarterly and yearly. Most of the researchers have used one of the prediction techniques in prediction of time series data. But, they have not tested all prediction techniques on same data set. They have not even compared the performance of different prediction techniques on the same data set. In this research work, some well known prediction techniques have been applied in the same time series data set. The average error and residual analysis have been done for each and every applied technique. One technique has been selected based on the minimum average error and residual analysis among the all applied techniques. The residual analysis comprises of absolute residual, maximum residual, median of absolute residual, mean of absolute residual and standard deviation. To finalize the algorithm, same procedure has been applied on different time series data sets. Finally, one technique has been selected which has been given minimum error and minimum value of residual analysis in most cases.
This Presentation is on recommended system on question paper predication using machine learning techniques. We did literature survey and implement using same technique.
2. A Review and Implementation of Principal Component Analysis 2
Abstract
In this experiment, we shall look at the famous iris data set and perform principal
component analysis on the data. We want to see which are the principal components that
explain the most variance within the data set. Furthermore, we will discuss the
application of principal component analysis in conjunction with other data analysis
techniques. All computations were performed in Python and all data is uploaded from the
UCI machine learning repository. Rather that simply using the built in PCA function, we
shall implement principal component analysis by manually performing each step, with
assistance from packages for Eigen-decomposition. In conclusion, we find that first two
principal components, out of four in total, explain roughly 96% of the variance in the data.
I. What is Principal Component Analysis?
Principal component analysis (PCA) is a orthogonal linear transformation of data, in
which the transformed data is projected onto a new coordinate plane. This transformed
data is displayed in such a manner that the first coordinate is the location of the greatest
variance, and every subsequent variance is placed on the coordinate system in a
decreasing fashion. These coordinates themselves are the principal components of the
data. The primary purpose of principal component analysis is “to reduce the
dimensionality of a data set consisting of a large number of interrelated variables, while
retaining as much as possible of the variation present in the data set.” (Wood, pg.2)
3. A Review and Implementation of Principal Component Analysis 3
II. Notation
𝑥 = 𝑣𝑒𝑐𝑡𝑜𝑟 𝑜𝑓 𝑝 𝑟𝑎𝑛𝑑𝑜𝑚 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 , 𝛼! = 𝑣𝑒𝑐𝑡𝑜𝑟 𝑜𝑓 𝑝 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡𝑠
𝛼!
!
𝑥 = Σ!
!
𝛼!" 𝑥!, Σ = 𝐶𝑜𝑣. 𝑚𝑎𝑡𝑟𝑖𝑥 𝑓𝑜𝑟 𝑥, 𝑟𝑒𝑝𝑙𝑎𝑐𝑒𝑑 𝑏𝑦 𝑆 𝑠𝑎𝑚𝑝𝑙𝑒 𝑐𝑜𝑣. 𝑚𝑎𝑡𝑟𝑖𝑥
𝜆! = 𝑃𝑟𝑖𝑛𝑐𝑖𝑝𝑎𝑙 𝐶𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡 𝑘, 𝑘 = (1,2, … , 𝑝)
III. Derivation of Principal Component Analysis
Our goal is to find the linear function of random variables from the x vector with the
vector of constants from the alpha vector with the maximum variance. This linear
function produces our principal components. Be this as it may, each principal component
must be in order of decreasing variance, and each principal component must be
uncorrelated with each other.
Objective:
𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒 𝑉𝑎𝑟 𝛼!
!
𝑥 = 𝛼!
!
Σ𝛼!
We seek to used constrained optimization, as without a constraint the value
of 𝛼! could be infinitely large. As such, we shall choose the following normalization
constraint:
𝛼!
!
𝛼! = 1
This brings us to the concept of Lagrange multipliers, which shall be the method
by which we achieve this constrained optimization.
4. A Review and Implementation of Principal Component Analysis 4
Lagrange Multipliers in PCA
The Lagrange Multiplier method is a tool “for constrained optimization of
differentiable functions, especially for nonlinear constrained optimization.”(Huijuan,
pg.1) In particular, this is helpful for finding local maxima and minima of a respective
function subject to a given constraint. Within the context of the experiment, the Lagrange
multipliers are applied as follows:
𝛼!
!
Σ𝛼! − 𝜆 𝛼!
𝛼! − 1
𝑑
𝑑𝛼!
𝛼!
!
Σ𝛼! − 𝜆 𝛼!
𝛼! − 1 = 0
Σ𝛼! − 𝜆𝛼! = 0
Σ𝛼! = 𝜆! 𝛼!
The final step of the equation yields us the eigenvector 𝛼! with its corresponding
eigenvalue 𝜆!.
What are Eigenvalues and Eigenvectors?
An eigenvalue is a number derived from a square matrix, which corresponds to a
specific eigenvector, also associated with a square matrix. Together, they “provide the
Eigen-decomposition of a matrix.” (Abdi, pg.1) Plainly spoken, the Eigen-decomposition
of a matrix merely provides the matrix in the form of eigenvectors and their
corresponding eigenvalues. Eigen-decomposition is important because it is a “method by
which we can find the maximum (or minimum) of functions involving matrices.” (Abdi,
pg.1) In this context, this is the method by which we find the principal components in
order of decreasing variance.
5. A Review and Implementation of Principal Component Analysis 5
Eigen-decomposition
𝐴𝑢 = 𝜆𝑢
𝐴 − 𝜆𝐼 𝑢 = 0
Where
A = square matrix,
u = eigenvector to matrix A (if length of vector changes when multiplied by A)
Assume that
𝐴 =
2 3
2 1
, 𝑇ℎ𝑒𝑟𝑒𝑟𝑓𝑜𝑟𝑒
𝑢! =
3
2
, 𝜆! = 4
𝑢! =
−1
1
, 𝜆! = −1
For most applications, the eigenvectors are normalized to a unit vector as such:
𝑢!
𝑢 = 1
Eigenvectors of A furthermore are put together together in a matrix U. each column of U
is an eigenvector of A. The eigenvalues are stored in a diagonal matrix ⋀, where the trace,
or diagonal, of the matrix gives the eigenvalues. Thus, we rewrite the first equation
accordingly:
𝐴𝑈 = 𝑈𝐴
𝐴 = 𝑈⋀𝑈!!
=
3 −1
2 1
4 0
0 −1
2 2
−4 6
=
2 3
2 1
6. A Review and Implementation of Principal Component Analysis 6
Moving forward, as we have mentioned prior, our objective is the maximize 𝜆!, and with
the eigenvectors defined in decreasing order. If 𝜆! is the largest eigenvector, then the first
principal component is defined as
Σ𝛼! = 𝜆! 𝛼!
In general, we define a given eigenvector as the k-th principal component of x and that
the variance of a given eigenvector is denoted by its corresponding eigenvalue. We shall
now demonstrate this process when k = 2 and when k > 2.
2nd
and K-th Principal Component
The second principal component maximizes the variance subject to being
uncorrelated with the first principal component The non-correlation constraint is
expressed as the following:
𝑐𝑜𝑣 𝛼!
!
𝑥𝛼!
!
𝑥 = 𝛼!
!
Σ𝛼! = 𝛼!
!
Σ𝛼! = 𝛼!
!
𝜆! 𝛼!
!
= 𝜆! 𝛼!
!
𝛼 = 𝜆! 𝛼!
!
𝛼! = 0
𝛼!
!
Σ𝛼! − 𝜆! 𝛼!
!
𝛼! − 1 − 𝜙𝛼!
!
𝛼!
𝑑
𝑑𝛼!
𝛼!
!
Σ𝛼! − 𝜆! 𝛼!
!
𝛼! − 1 − 𝜙𝛼!
!
𝛼! = 0
= Σ𝛼! − 𝜆! 𝛼! − 𝜙𝛼! = 0
𝛼!
!
Σ𝛼! − 𝛼!
!
𝜆! 𝛼! − 𝛼!
!
𝜙𝛼! = 0
0 − 0 − 𝜙1 = 0
𝜙 = 0
Σ𝛼! − 𝜆! 𝛼! = 0
This process can be repeated up to k = p, yielding principal components for each
of the p random variables.
7. A Review and Implementation of Principal Component Analysis 7
IV. Data
For this experiment, we shall be using Ronald Fisher’s Iris flower data set, originally
collected by Edgar Anderson to study the variation of the three species. Our objective is
to determine which principal components contain the most data regarding this data set.
There are a total of 150 observations, 50 of each of the three species of flower. The
species and variables of observed are:
Species
• Iris-Setosa
• Iris-Virginica
• Iris-Veriscolor
Variables
• Sepal Length
• Sepal Width
• Petal Length
• Petal Width
8. A Review and Implementation of Principal Component Analysis 8
V. Experiment
When performing initial exploratory analysis on our data, we notice the following:
As we observe, the data exhibits very high variance within and between species
with respect to sepal length and sepal width, but is considerable less variable between
species, and moderately variable within species when observing petal length and petal
width. This will be a point of interest to keep in mind for later, but for now let us move
on to describing the implementation as performed here. After we load our data into a
variable within Python, we standardize our values (mean = 0, var. =1), then we calculate
the covariance matrix for X:
𝑆 = Σ!
!
𝑥! − 𝑥 !
(𝑥 − 𝑥)
9. A Review and Implementation of Principal Component Analysis 9
Generally speaking, we want to standardize values when they are not measured on
the same scale. Although in this experiment all of the variables are measured on a
centimeter scale, it is advisable to still do so. Moving forward, we perform the Eigen-
decomposition, and obtain the eigenvalues and eigenvectors. After we sort the
eigenvectors, we observe the following:
As we can see, the first two principal components explain the vast majority of
variance within the data set. As pointed out early, the high variability amongst sepal
length and sepal width between and within species foreshadowed these events. Finally,
we project the transformed data onto the new feature space:
10. A Review and Implementation of Principal Component Analysis 10
VI. Conclusion and Comments
We observe that instead of a 4-dimensional plot, as we would have originally had,
we are now looking at a very familiar xy plot. For exploratory analysis purposes, this
brings considerable ease both visually and analytically. It is easy to see that Iris-viriginica
and Iris-veriscolor show considerable similarities with respect to sepal length and sepal
width properties. In contrast, Iris-setosa in general seems to be considerably unique. As
for further applications of PCA, it is used in regression analysis often to determine which
variables should be included in a model, used in neuroscience to identify properties of
stimuli, as well as other tools. As proven above, both in theory and application, principal
component analysis provides a robust and excellent method of simplifying very complex
data into less complex forms.
11. A Review and Implementation of Principal Component Analysis 11
VII. Appendix
1. Wood, F. (2009, September). Principal Component Analysis. Retrieved from
http://www.stat.columbia.edu/~fwood/Teaching/w4315/Fall2009/pca.pdf
2. Abdi, H. (2007). The Eigen-Decomposition. Retrieved from
https://www.utdallas.edu/~herve/Abdi-EVD2007-pretty.pdf
3. Huijuan, L. (2008, September 28). Lagrange Multipliers and their
Applications. Retrieved from
http://sces.phys.utk.edu/~moreo/mm08/method_HLi.pdf