Time Series Analysis and Forecasting

Lesson 01: Introduction

by
Dr. Rajat Kanti Samal
Assistant Professor

Department of Electrical Engineering

Veer Surendra Sai University of Technology
Burla, Odisha – 768018

Dr. R. K. Samal (EE) Lesson 01 Dec 2022 1 / 11

Overview

Books
1 Time Series Analysis and Its Applications by Robert H. Shumway
David S. Stoffer.
2 Time Series Analysis by Jonathan D. Cryer, Kung-Sik Chan.
Prerequisites
1 Basics of Statistics
2 Basics of Linear Algebra
Course Outcomes (CO)
1 Exploratory data analysis
2 ARIMA Models (time domain)
3 Spectral analysis (frequency domain)
4 GARCH and ARMAX models
5 State space models
Further courses
1 Data Science and Analytics
2 Statistical and Machine Learning.

Dr. R. K. Samal (EE) Lesson 01 Dec 2022 2 / 11

Statistics and Data Science

Dr. R. K. Samal (EE) Lesson 01 Dec 2022 3 / 11

Statistics and Computer Science

Statistics, data mining, and machine learning are all concerned with
collecting and analyzing data.
Statisticians thought that computer scientists were reinventing the
wheel. Computer scientists thought that statistical theory did not ap-
ply to their problems.
Things have changed. Statisticians now recognize that computer sci-
entists are making novel contributions while computer scientists now
recognize the generality of statistical theory and methodology.
Clever data mining algorithms are more scalable than statisticians ever
thought possible. Formal statistical theory is more pervasive than com-
puter scientists have realized.
Students who analyze data, or who aspire to develop new methods
for analyzing data, should be well grounded in basic probability and
mathematical statistics.

Dr. R. K. Samal (EE) Lesson 01 Dec 2022 4 / 11

Probability and Statistical Inference
Probability is more engaging when students can see it put to work in
the context of statistics.
Probability is the formal language of uncertainty which is the basis
for statistical inference. The basic problem we study is Given a data
generating process, what are the properties of the outcomes?
The basic problem is statistical inference is Given the outcomes, what
can we say about the process that generated the data?.

Dr. R. K. Samal (EE) Lesson 01 Dec 2022 5 / 11

Probability and Inference (cont.)

Prediction, classification, clustering and estimation are all special cases

of statistical inference.
Data analysis, machine learning and data mining are various names
given to the practice of statistical inference depending on context.
Statistical inference can be used for specific problems such as regres-
sion, grahical models, causation, density estimation, smoothing, clas-
sification and simulation.
Stochastic processes including Markov chains.
Statisticians and computer scientists use different language for the same
thing, for example
estimation/learning
classification/supervised learning
clustering/unsupervised learning
data/training sample; classifier/hypothesis

Dr. R. K. Samal (EE) Lesson 01 Dec 2022 6 / 11

Time Series Analysis

The analysis of experimental data recorded at different points in time

includes correlation between adjacent data points.
This correlation is not addressed by conventional statistical models
which assume that adjacent observations are independent and identi-
cally distributed.
The systematic approach which is used for analyzing data recorded at
different points in time have correlation between adjacent observations
is known as time series analysis.
The primary objective of the time series analysis is to develop mathe-
matical models that provide plausible descriptions for the sample data.
The first step in any time series investigation always involves careful
examination of recorded data plotted over time.
The scrutiny often suggests methods of analysis and statistics that will
be of use in summarizing the information in the data.

Dr. R. K. Samal (EE) Lesson 01 Dec 2022 7 / 11

Time Series Analysis (cont.)

1 Two separate (not mutually exclusive) approaches to time series anal-

ysis exist:
time domain approach (lagged relationships)
frequency domain approach (investigation of cycles)
2 The time domain approach views the investigation of lagged relation-
ship as most important e.g. how does what happened today will affect
what will happen tomorrow.
3 The frequency domain approach views the investigation of cycles as
most important e.g. what is the economic cycle through periods of
expansion and recession.
4 The problems and questions of interest to the prospective time series
analyst can be exposed by considering real experimental data taken
from different subject areas.

Dr. R. K. Samal (EE) Lesson 01 Dec 2022 8 / 11

Some Applications of Time Series Analysis

Field of economics as in daily stock market quotations or monthly

employment figures.
Social scientists follow population series such as birth rates or school
enrollments.
an epidemologist might be interested in number of influenza cases ob-
served over a time period.
In medicine, blood pressure measurements traced over time could be
useful for evaluating drugs used in treating hypertension.
Functional magnetic resonance imaging of brain wave time series pat-
terns might be used to study how brain reacts to certain stimuli under
various experimental conditions.
Wind speed and solar irradiance time series for renewable power.
Electrical load time series.

Dr. R. K. Samal (EE) Lesson 01 Dec 2022 9 / 11

The R programming language

1 R is a statistical computer program available under General Public

License (GPL). Thus it can be freely distributed.
2 R is rich in tools and packages to perform statistical analysis and
produce graphics. It is in fact a complete programming language.
3 Inspired by the C nomenclature, Becker and Chambers (1980s) called
their newly developed statistical programming language S.
4 Ross Ihaka and Robert Gentleman from the University of Auckland,
New Zealand, wrote a reduced version of S for teaching purposes.
5 In 1995, Martin Maechler persuaded Ross and Robert to release the
source code for R under the GPL.
6 This coincided with the upsurge in Open Source software spurred by
the Linux system.
7 Some issues: packages may not work in higher versions.
8 Assignment: Install R-base, RStudio and the astsa package.
Dr. R. K. Samal (EE) Lesson 01 Dec 2022 10 / 11
In God we trust, all others bring data.

Dr. R. K. Samal (EE) Lesson 01 Dec 2022 11 / 11