univariate and bivariate analysis.
Uploaded by
ScribdTranslationsunivariate and bivariate analysis.
Uploaded by
ScribdTranslationsUnivariate and Bivariate Analysis
● Members of the Matricula team
● Jessica Ivett Morua Enriquez 20M07C05
● Claudia Hernandez Chavez 20M07B78
● Ivan Gerardo Cervantes Esparza 20M07B61
● Nestor Daniel Moreno Gonzalez 20M07C02
● Maria Dolores Garcia Hernandez 20M07B69
● Carlos Rosas Jimenez 20M07C26
● Eduardo Torres Grimaldo 20M07C37
● Jose Luis Palomo Juarez 20M07C15
● Juan Gabriel Suarez Loredo 20M07C36
Module: Statistics
Professor: Julio Ferretiz Aguilar
Date: August 7, 2021
Team 1
STATISTICAL ANALYSIS Statistical analysis is divided into three main types: Univariate
Bivariate Multivariate
Univariate Analysis.
Univariate analysis allows us to analyze and understand the statistical processing
of a variable, whose objective is:
“Analyze and synthesize the information contained in statistical data using: tables,
graphs and numerical summaries.
Univariate analysis aims to form an idea of the specific characteristics of the
problem; not in a general way, but rather variable by variable. Two classes of
parameters were included in this study:
a) Central tendency (mean, mode, median).
b) Dispersion (standard deviation, variance, correlation coefficient).
With these two types of values we will know the behavior or tendency of each
variable to make a subsequent decision about which are the most appropriate
techniques for multivariate analysis.
What is a univariate analysis?
It consists of the analysis of each of the variables studied separately, that is, the
analysis is based on a single variable. The most frequent techniques of univariate
analysis are the frequency distribution for a univariate table and the analysis of the
central tendency measures of the variable. It is used only in those variables that
were measured at the interval or ratio level (see Therese L. Baker, 1997). The
frequency distribution of the variable requires seeing how the categories of the
variable are distributed, and can be presented based on the number of cases or in
percentage terms.
How to present the results and draw a conclusion about them is possibly the most
important part of statistics. Remember that a good report "is one that presents
statistically reliable data, with understandable tables or graphs."
Example:
Number of US Printing Industry Companies by Segment
Where:
µ = population mean
ΣXi = Summation of the scores
N = Number of cases
Example in everyday life:
You weigh the puppies and get these results:
2,5, 3,5, 3,3, 3,1, 2,6, 3,6, 2,4
The variable is the weight of the puppy
What is a univariate analysis?
As seen, univariate analysis methods are used to study the behavior of variables
individually. The main advantage of frequency distributions is that they can be
carried out regardless of the scale on which the variables to be analyzed are
measured.
Bivariate Analysis
Bivariate data are those that come from two variables measured at the same time
on each individual. For example: Age and Gender, Education and Income, Weight
and Height, etc.
When bivariate data comes from two quantitative variables, it is of interest to study
the relationship between one and the other. The relationship can be of a very
different nature: linear, quadratic, exponential, logarithmic, trigonometric, etc. In
statistics, the relationship that interests us is the Linear Relationship, which is why
Linear Correlation and Linear Regression Analysis are carried out. Correlation
analysis is used to measure the strength of association between variables. The
objective is to measure the covariance that exists between these two numerical
variables. Regression analysis is used for prediction purposes. The aim is to
develop a statistical model useful for predicting the values of a dependent or
response variable based on the values of at least one independent or explanatory
variable.
With bivariate analysis we describe and evaluate the relationship between two
variables or the behavior of one variable as a function of another. Through
multivariate analysis we study three or more variables simultaneously and try to
determine the relationship between them.
What is data analysis, differentiating between univariate, bivariate and multivariate
data analysis?
This is quantitative data analysis. Analyses of this type can be univariate, that is,
assess a single variable, bivariate when we focus on finding correlations between
two variables, or multivariate, which consists of studying more than two variables.
Bivariate analysis techniques:
● Analysis for categorical data: Pearson Chi-square, Fisher's Exact Test,
McNemar Test for paired data.
● Analysis for quantitative data: Pearson correlation, Spearman correlation,
Kendall's Tau.
● Comparison between groups of independent and paired samples: Student's
t test, analysis of variance (ANOVA), Mann-Whitney test, median test,
Kruskal-Wallis test, Wilcoxon test, Friedman test, robust methods.
● Comparison of analytical methods: Intraclass correlation coefficient (ICC),
Kappa index, Bland-Altman graphic method.
● Diagnostic techniques: determination of the precision of a method
(accuracy), ROC curves, determination of sensitivity and specificity.
Multivariate analysis techniques:
● Generalized linear models (GLM): Multiple linear regression, multiple logistic
regression, negative binomial regression, Poisson regression.
● Classification problems: discriminant analysis, factor analysis.
● Dimensionality reduction: principal components (PCA), correspondence
analysis, classification and regression trees (CART), Random Forests (RF).
● Cluster analysis.
Example :
It was decided to examine the relationship between height (in meters) and weight
(in kilograms) using a sample of 12 female students from a certain school. The
data is shown in the following table:
The dividing line between a univariate and a bivariate study lies in the fact that the
latter strictly seeks to analyze two variables together, which we could well denote
as “X” and “Y” seeking to prove simple relationships of causality or association.
The basic goal of bivariate analysis is to investigate the relationship between two
sets of data, such as pairs of observations taken from the same sample or
individual. In statistics, a specific variant of bivariate studies is called simple
“correlational analysis,” which seeks to demonstrate that there is a significant direct
or inverse relationship between two variables.
In simple words, bivariate studies measure the relationship between two variables
(bi means two). In most cases of two-variable studies, the scope is limited to
knowing whether variables “X” and “Y” are linked. A coefficient or statistical ratio
that measures the degree of correlation is the coefficient of determination or “R
squared.”
For example, if you want to see how family income influences graduation rates,
you can use a bivariate correlation to examine the two variables.
Example in everyday life:
A local ice cream shop tracks how much ice cream they sell compared to
the temperature that day.
The two variables are Ice Cream Sales and Temperature.
Conclusion
Univariate analysis
It seeks to understand the behavior of each of the dependent variables analyzed
both in its original form and in the dichotomous form, this will allow us to
understand the way in which the original variables were transformed to construct
the dependent variables that allow the use of subsequent analysis techniques and
at the same time understand the distribution behavior of the same.
Bivariate analysis
It is the one that allows to identify more clearly the causal behaviors that may exist
between the dependent and explanatory variables since it consists of generating
the statistics that come from the intersection between these two types of variables
according to their role.
A descriptive video of the bivariate analysis is left
https://youtu.be/tKY0sjmA1Fk