Abstract
The prediction models in machine learning and the application of different algorithms help in developing appropriate knowledge of future behaviour based on past data. With a view to help developing insights, this chapter explains the basics of predictive analytics with its common traits. The main objective of this chapter and the topics covered in the predictive models show how different methods of tools and algorithms are crucial for predictive analytics and steps to be used to make a good model. It also describes the basic tools and techniques to be used in predictive modelling. The discussions are based on the data of live projects in predictive modelling. It attempts to handle the predictive analytics like linear regression, multivariate linear regression, nonlinear regression, and multivariate non-linear regression. These statistical tools are crucial and critical in future technology developments.
TopIntroduction
Technology development requires accurate calculation and computation of the data that are required for the correct usage and application. This is possible only when a modelling is tried and tested before launching the technology for real time usage. A close analysis of the past data can only help in understanding the possible behaviour of the future events. Predictive modelling is the fundamental sine qua non for future technology developments and this needs predictive analytics. Predictive modelling and predictive analytics are terms which are interchangeably used (Joseph M. Carew, 2020), In predictive modelling, a mathematical process is involved leading to the prediction of future events or outcomes by analysing past patters of the data. It is always endeavoured to explore the most likely eventuality based on the past behaviour of various data. This requires data collection which has to be carried out selectively and meticulously. Once the data is available, the analyst can use statistical models using the historical data. Due care is to be taken to avoid data errors like the Alpha Error or Beta Error. According to Statistical theorems, large sample data may not be improving the accuracy of prediction to the Beta Error. It is said that “ All models are wrong, but some are useful” (Joseph M. Carew, 2020),
Availability of data
How does Regression works -Regression Analysis is a statistical process for estimating the relationship among variables. IT includes techniques for analyzing relationship between several variables, that is between a dependent variable and one or more independent variables. That means regression helps one understand how a value of the dependent variable varies when anyone of the independent variables is changed, while the other independent variables remain fixed. In most cases, regression estimates the average value of the dependent variable when the independent value doesn’t change.
In some cases, the focus is on the other location parameter of the conditional distribution of the dependent variable given the independent variables. The goal of estimation is to obtain a function of the independent variables, called the regression function, which can be linear or nonlinear. To characterize the variation of the dependent variable around the regression function is also an agenda of regression analysis, which can be described by a probability distribution. In regression analysis, to best fit a set of data observations that you provide, the values of parameters for a function are to be determined. Equation expresses these relationships in symbols. It shows that regression is the process of estimating the value of a continuous target (y) as a function (F) of one or more predictors (x1, x, an), a set of parameters (01, 0,, O), and a measure of error (e).
Y=f(x,O)+e
Multivariate Linear Regression
The term multivariate linear regression refers to linear regression with two or more predictors (x1, x2, …., xn). The regression line cannot be visualized in two-dimensional space, when multiple predictors are used; still the line can be computed by expanding the equation for linear regression to include the parameters for each of the predictor variables.
Y = ɵ
1 + ɵ
2 x
1 + ɵ
3 x
2+ … + ɵ
nx
n – 1 + e
Regression coefficients: The regression parameters are often referred to as coefficients in multivariate linear regression. In this case, the algorithm computes a coefficient for each of the predictors used by the model. The impact of the predictor x on the target y is given by the coefficient of it. To analyze the regression coefficients and to evaluate how much the regression line fits the data, many statistics such as R-squared value are available.
Figure 1.
Nonlinear aggression with a single predictor
Key Terms in this Chapter
Artificial Intelligence: Artificial intelligence is intelligence demonstrated by machines, unlike the natural intelligence displayed by humans and animals, which involves consciousness and emotionality. The distinction between the former and the latter categories is often revealed by the acronym chosen.
Predictive Analytics: Predictive analytics encompasses a variety of statistical techniques from data mining, predictive modelling, and machine learning that analyze current and historical facts to make predictions about future or otherwise unknown events.
Beta Error: The statistical error (said to be 'of the second kind,' or type II) that is made in testing when it is concluded that something is negative when it really is positive.
Regression: Regression analysis is a statistical method that helps in analysing and understanding the relationship between two or more variables of interest.
Statistical model: A statistical model is a mathematical model that embodies a set of statistical assumptions concerning the generation of sample data. A statistical model represents, often in considerably idealized form, the data-generating process.
Alpha Error: The statistical error made in testing a hypothesis when it is concluded that a result is positive, but it really is not.
Small Sample: If the sample size n ils less than 30 (n<30), it is known as small sample. For small samples, the sampling distributions are t, F and ?2 distribution.