Page 42 - IJTSRD.com - Special Issue - 2nd International Congress of Engineering

International Journal of Trend in Scientific Research and Development (IJTSRD) ISSN: 2456-6470
nd
Special Issue of 2 International Congress of Engineering
by two criteria: the quantitative and the qualitative. Following the ideas of the same authors, they mention
The former analyzes historical data using that in this model, y is a linear function of x (the part
mathematical models and statistics, and the second α + βx) plus ε (Greek letter epsilon) representing the
using knowledge of the current market situation and error y is a random variable. Anderson, Sweeney and
its environment. The best production forecast will be Williams (2001) point out that the error term explains
the one with the best information mix of both criteria. the variability in y that cannot be explained by the
Chapman (2006) defines the formulation of forecasts linear relationship.
of the technique to use past experiences in order to
predict expectations of the future. The method of least squares has a long history that
goes back to the beginning of the nineteenth century.
Montgomery, D., Peck, E. and Vining (2006) mention In June 1801, Zach, an astronomer Gauss had known
that linear regression models are widely used in two years earlier, published the orbital positions of the
engineering since they serve to analyze the behavior celestial body Ceres, a new "small planet" discovered
of input (or regressor) and output (or response) by the Italian astronomer G. Piazzi in the same year.
variables predictions and estimates. On the other Unfortunately, Piazzi had only been able to observe 9
hand, Badii, Guillen, Cerna, Valenzuela and Landeros degrees of its orbit before this body disappeared after
(2012) indicate that regression and correlation are two the sun. Zach published several predictions of his
closely related techniques and comprise a form of position including one of Gauss that differed
estimation. More specifically, correlation and remarkably from the others. When Ceres was
regression analysis include the study of sampling data rediscovered by Zach in December 1801 it was almost
to know what two or more variables are related to one exactly where Gauss had predicted (Cruces, S / A).
another in a population. Correlation analysis produces The method of ordinary least squares consists of
a number that summarizes the degree of correlation obtaining a hyperplane so that the sum of the squares
between two variables; and regression analysis gives of the distances between each of the observations of
rise to a mathematical equation that explains and the variable and said hyperplane (residues)
predicts this relationship. (Chirivella, S/A).

Lopez and Romero (2014) mention that the simple or GENERAL OBJECTIVE
bivariate RL models are used as models of prediction
or prognosis. The most typical case is when the Assess the quality of a prognosis for an industrial
predictor, regressor or independent variable X is a product using the regression analysis.
controlled variable (non-random), while the response
variable or dependent variable Y is a random variable Specific objectives
that has an approximately normal distribution for each  Understand the contextualization of the topic.
2
x value of X, but with constant variance  .
Escalante (2013) mentions that regression analysis is  Know the existing models to evaluate the quality
a technique used to relate, through a model, one or of an industrial process.
more independent variables to a dependent variable  Calculate the representative sample using 95%
(response). confidence.

Orellana (2008) mentions that the simplest function  Apply the regression analysis on the
for the relationship between two variables is the linear representative samples.
function:  Analyze the results obtained from the regression

Y = a + b X analysis on the representative samples.
 Evaluate the results obtained from the regression
Cardona, González, Rivera and Cárdenas (2013) analysis on the representative samples.
mention that the general equation describing the
relationship between the two variables is:

y = α + βx + ϵ

@ IJTSRD | Available Online @ www.ijtsrd.com | Special Issue Publication | November 2017 P - 20