Book Image

Mathematica Data Analysis

By : Sergiy Suchok
Book Image

Mathematica Data Analysis

By: Sergiy Suchok

Overview of this book

There are many algorithms for data analysis and it’s not always possible to quickly choose the best one for each case. Implementation of the algorithms takes a lot of time. With the help of Mathematica, you can quickly get a result from the use of a particular method, because this system contains almost all the known algorithms for data analysis. If you are not a programmer but you need to analyze data, this book will show you the capabilities of Mathematica when just few strings of intelligible code help to solve huge tasks from statistical issues to pattern recognition. If you're a programmer, with the help of this book, you will learn how to use the library of algorithms implemented in Mathematica in your programs, as well as how to write algorithm testing procedure. With each chapter, you'll be more immersed in the special world of Mathematica. Along with intuitive queries for data processing, we will highlight the nuances and features of this system, allowing you to build effective analysis systems. With the help of this book, you will learn how to optimize the computations by combining your libraries with the Mathematica kernel.
Table of Contents (10 chapters)
9
Index

Choosing the best time series process model


As we have become acquainted with the main processes, now we can proceed with their forecasts. To do this, let's get to know the TimeSeriesModelFit function by considering the following example. Suppose we have data on the air temperature in London from January 1, 2012 to September 1, 2015, and we want to get the forecast for the next 6 days. Moreover, we want to compare how the forecast we receive will differ from the actual one. We will also build 95% confidence intervals that should include almost all the possible results of the forecast:

Take into account that Mathematica has reported that the SARIMA model is the most suitable one for this time series. This is true, as air temperature is highly dependent on season. Further more, in order to build a graph, we did not take all the data, but just the part from August, by cutting off all other data using the TimeSeriesWindow function. In order to build a forecast, we called the TimeSeriesForecast...