1 of 50

��������������������Lecture 1 � �Introduction to Time Series Analysis & Forecasting �

1

Instructor: Ercan Atam

Institute for Data Science & Artificial Intelligence

​

Course: DSAI 525-Time Series and Forecasting with Machine Learning

2 of 50

2

List of contents for this lecture

​

​

  • What is a time series?

​

  • What is forecasting?

​

  • Examples of time series

​

  • Introduction of terminology related to time series and forecasting

​

  • Components of a forecasting process

​

  • Data concepts and some data processing methods

​

​

3 of 50

3

What is forecasting?

4 of 50

4

Forecasting applications

5 of 50

5

Forecasting ranges

6 of 50

6

Backbone of forecasting: time series (1)

7 of 50

7

​

Backbone of forecasting: time series (2)

​

8 of 50

8

Why is forecasting so important? (1)

9 of 50

9

Why is forecasting so important? (2)

10 of 50

10

​

Why is forecasting so important? (3)

​

11 of 50

11

​

Why is forecasting so important? (4)

​

12 of 50

12

Broad categorization of forecasting methods

13 of 50

13

Delphi method

14 of 50

14

​

Quantitative forecasting methods

​

15 of 50

15

Terminology (1)

Point forecast (point estimate)

​

16 of 50

16

Terminology (2)

Prediction interval

​

​

17 of 50

17

Terminology (3)

Forecast horizon (lead time)

​

​

18 of 50

18

Terminology (4)

Forecasting interval

​

​

19 of 50

19

Terminology (5)

Moving horizon forecasts

 

 

 

 

20 of 50

20

Examples of time series (1)

Uncorrelated data, constant process model

FIGURE 1.5. Pharmaceutical product sales

21 of 50

21

Examples of time series (2)

Autocorrelated data

FIGURE 1.6. Chemical process viscosity readings

22 of 50

22

Examples of time series (3)

Trend

FIGURE 1.7. The U.S. annual production of blue and gorgonzola cheeses. (Source: USDA-NASS.)

23 of 50

23

Examples of time series (4)

Cyclic or seasonal data

FIGURE 1.8. The U.S. beverage manufacturer monthly product shipments, unadjusted. (Source: U.S. Census Bureau.)

24 of 50

24

Examples of time series (5)

Non-stationary data

FIGURE 1.9. Global mean surface air temperature anomaly. (Source: NASA-GISS.)

25 of 50

25

Examples of time series (6)

A mixture of patterns

FIGURE 1.11. Monthly unemployment rate – full-time labor force, unadjusted. (Source: U.S. Department of Labor – BLS.)

26 of 50

26

Examples of time series (7)

Cyclic patterns of different magnitudes

FIGURE 1.12. The international sunspot number. (Source: SIDC)

27 of 50

27

Examples of time series (8)

A typical events

FIGURE 1.13. Pharmaceutical product sales.

FIGURE 1.14. Chemical process viscosity readings, with sensor malfunction.

28 of 50

28

The forecasting process

29 of 50

29

Data for forecasting (1)

The Data Warehouse

​

30 of 50

30

Data for forecasting (2)

The Data Warehouse (continue)

​

31 of 50

31

Data for forecasting (3)

Data versus information

​

32 of 50

32

​

Data for forecasting (4)

​

Data extraction and wrangling

​

33 of 50

33

​

Data for forecasting (5)

​

Data transformation

​

34 of 50

34

​

Data for forecasting (6)

​

Data cleaning

​

35 of 50

35

Data quality (1)

36 of 50

36

​

Data quality (2)

​

Accuracy

37 of 50

37

​

Data quality (3)

​

Timeliness

38 of 50

38

​

Data quality (4)

​

Completeness

39 of 50

39

​

Data quality (5)

​

Representativeness

40 of 50

40

​

Data quality (6)

​

Consistency

41 of 50

41

​

Data imputation (1)

​

42 of 50

42

​

Data imputation (2)

​

Mean value imputation

43 of 50

43

​

Data imputation (3)

​

Stochastic mean value imputation

44 of 50

44

​

Data imputation (4)

​

Mean value imputation using a subset of data

45 of 50

45

​

Data imputation (5)

​

Regression imputation

46 of 50

46

​

Data imputation (6)

​

Hot deck imputation

47 of 50

47

​

Data imputation (7)

​

Hot deck imputation (continue)

48 of 50

48

​

Data imputation (8)

​

Cold deck imputation

49 of 50

49

​

Data imputation (9)

​

Cold deck imputation (continue)

50 of 50

References �(utilized for preparation of lecture notes or MATLAB code)

50

​

  • “Introduction to Time Series Analysis and Forecasting”, Douglas C. Montgomery, Cheryl Jennings, Murat Kulahci. Wiley, 2024.

​