For each layer there is typically a one- to- one relationship between geographic features (point, line, or polygon) and records in a table

Transcription

1 For each layer there is typically a one- to- one relationship between geographic features (point, line, or polygon) and records in a table

2 Common components of a database: Attribute (or item or field) Record (or tuple) integer domain real domain alpha- numeric domain (a string)

3 A DBMS provides Translation (many views on the data) Protection (e.g., against errors due to simultaneous updates)

4 Multi- tiered architecture

5 Disk Farms of Storage 5

6 Common Database Models: Flat File Hierarchical Network Relational

7 Flat File Little cross- referencing, often row/column format, ransparent, easily transportable, but redundant, wastes space, less flexible, slow processing, few error safeguards

8 Relational Model Minimal row- column structure Items/records with specified domains (possible values) Advantages: Minimum structure, easy programming, flexible Disadvantages: Relatively slow, a few restrictions on attribute content

9 Hybrid Database Coordinate data as a connected structure Attribute data as a relational table

10 Eight Fundamental Operations Restrict (query) subset by rows Project subset by columns Product all possible combinations Divide inverse of product

11 Eight Fundamental Operations Union combine top to bottom Intersect row overlap Difference row non- overlap Join (relate) combine by a key column

12 Main Operations with Relational Tables Query / Restrict Conditional selection Calculation and Assignment Sort rank based on attributes Relate/Join Temporarily combine two tables by an index

13 Query / Restrict Operations with Relational Tables Set Algebra Operation less than (<), greater than (>), equal to (=), and not equal to (<>), and others Boolean Algebra uses the conditions OR, AND, and NOT to select features. Boolean expressions are evaluated by assigning an outcome, True or False, to each condition.

14 Query Animal = cat

15 Query: AND results in reduced set Animal = cat AND Stance = sit

16 Query: OR increases selected set Animal = cat or Stance = stand

17 Animal NOT cat

18 Query / Restrict simple, AND

19 Query / Restrict OR, NOT

20 Operation Order is Important in Query (D OR E) AND F may not be the same as D OR (E AND F) NOT (A and B) may not be the same as [ NOT (A) AND NOT (B)] Typically need to clarify order with delimiters

21 21

22 22

23 23

24 24

25 25

26 26

27 27

28 28

29 29

30 30

31 31

32 32

33 Structured Query Language (SQL) A standard system for query syntax Uniformly interpreted set of operations, e.g., CREATE INSERT SELECT Anybody can (and it appear everybody has) create a database engine that fits under SQL then we can switch vendors and upgrade whenever we want.in theory.

34 SQL Example CREATE TABLE informix.culverts ( se_row_id serial primary key, shape ST_Point); CREATE INDEX culverts_ix2 ON culverts (shape ST_Geometry_Ops) USING RTREE; INSERT INTO geometry_columns VALUES ( "hamlet", "informix", "culverts",... Source: "shape", NULL, 1, -- The code for a Point. NULL, 0 );

35 Main Operations with Relational Tables Query / Selection Conditional selection Calculation and Assignment Sort rank based on attributes Relate/Join Temporarily combine two tables by an index

36 Calculation and Assignment Slope = steep Aspect = 45.2 Cost = [ M * U + cos (distance) ] / (F P/R*T)

37 Main Operations with Relational Tables Query / Selection Calculation and Assignment Sort rank based on attributes Relate/Join Temporarily combine two tables by an index

38 Sort ordering by attribute values Simple sort ascending AREA Compound sort ascending Type, then descending AREA within Type

39 Main Operations with Relational Tables Query / Selection Conditional selection Calculation and Assignment Sort rank based on attributes Relate/Join Temporarily combine two tables by an index

40 Tables in GIS Attribute tables are often huge We have to maintain our tables (change values, remove, add records or items) Different people/applications are interested in different subsets of attributes (columns) We often break our tables up into pieces (many tables), and use relational joins as needed to combine them back together

41 A Relational Join

42 Joins Are Through Keys Columns whose values uniquely define rows

43 Joins Are Through Keys Columns whose values uniquely define rows

44

45 Well- behaved join 45

46 46

47 A hiccup 47

48 Bad Foreign Key 48

49 Incomplete Match outer join inner join 49

50 Things to Remember About Relates: You need a key in both tables One value (source key) - to- many (target key) relates work well Many- to- one often don t work Missing values in the source typically result in null entries in the joined table Relates are usually virtual - the base tables don t change, they re just shown as combined tables 50

51 Relational Tables Relational tables have many advantages, but If improperly structured, table may suffer from: Poor performance Inconsistency Redundancy Difficult maintenance This is common because most users do not understand the concepts of Normal Forms in relational tables.

52 Tables in Non- normal Form repeat columns, dependent data, empty cells by design

53 Normal Forms Are Good Because: It reduces total data storage Changing values in the database is easier It insulates information it is easier to retain important data Many operations are easier to code

54 Why Should You Care? Public Data Structured This Way

55 Why Should You Care? Public Data Structured This Way 55

56

57 Keys - column(s) that uniquely identifies every row in a table candidate key - there may be more than one key in a table, each is called a candidate key primary key - the one chosen as the key for the table Functional dependency - if you know the value of column(s) A, then you automatically know the value of column B - this means column B is functionally dependent on column(s) A Every column in a table is functionally dependent on the primary key in the table Why are keys important? It s how we join tables, relating from a key

58 Normal forms is the process of breaking the information in table(s) up such that the data are in naturally grouped in tables and Each table has a key, such that tables can be related back to each other Typically we put tables into what is called the Third Normal Form

59 Keys Item(s) that uniquely identify a row STATE can be a key, but not REGION, SIZE, or POPULATION

60 What are the functional dependencies? What are the candidate keys? 80 Small Large 80 70

61 functional dependencies? Type - > Class 80 Small Large 80 70

62 functional dependencies? Type - > Size 80 Small Large 80 70

63 functional dependencies? Size - > Type 80 Small Large 80 70

64 functional dependencies? Type - > Shape 80 Small Large 80 70

65 functional dependencies? Length - > Shape 80 Small Large 80 70

66 functional dependencies? Length - > Class 80 Small Large 80 70

67

68 Keys - Items that uniquely identify the rows What single Items can be keys? 68

69 Keys - Items that uniquely identify the rows What PAIR of Items can be keys? PID+ Osel+ Clr+ NumT+ SpLM+ 69

70 Keys - Items that uniquely identify the rows What PAIR of Items can be keys? PID+ nothing, NULL Osel+NumT Clr+NumT NumT+SpLM, +Clr, +Osel SpLM+NumT 70

71 Tables in 1st normal form when there are no repeat columns Advantages: easy to code queries (can look in only one column) Disadvantages: slow searches, excess storage, cumbersome maintenance

72 2 nd Normal Forms in Relational Tables 2NF if: it is in 1NF and if every non- key attribute is functionally dependent on the primary key Remember, a key is an item or set of items that may be used to uniquely identify every row Remember, a functional dependency means if you know an item (or items) for a row, then you automatically know a second set of items for the row this means the second set of items is functionally dependent on the item (or items)

73 Keys Item(s) that uniquely identify a row Sometimes we need >1 column to form a key, e.g., Parcel- ID and Own- ID together may form a key

74 Functional Dependency Knowing the value of an item (or items) means you know the values of other items in the row e.g., if we know the person s name, then we know the address In our example, if we know the Parcel- ID, we know the Alderman, Township name, and other Township attributes: Parcel- ID - > Alderman Parcel- ID - > Thall_add Parcel- ID - > Tship- ID Parcel- ID - > Tship_name

75 Moving from First Normal Form (1NF to Second Normal Form (2NF), we need to: Identify functional dependencies Place in separate tables, one key per table

76 3 rd Normal Forms in Relational Tables Remove transitive functional dependencies A transitive functional dependency is when A - > B (if we know A, then we know B) and B - > C (if we know B, then we know C) So A - > C (if we know A, then we know C). To be in 3NF, we must identify all transitive functional dependencies, and remove them, typically by splitting the table(s) that contain them

77

78 In our example, one transitive functional dependency: Parcel- ID - > Tship- ID, Alderman Tship- ID - > Tship_name, Thall_add

79

80 Bad Things in Relational Tables: Repeat (or similar) variables e.g., parcel #, owner 1, owner2, owner3, owner 4 Multiple dependencies per record e.g., owner name, house#, street, city, county, zipcode, state, country Repeat records Many blank cells

81 Normal Forms Summary A) No repeat columns (create new records such that there are multiple records per entry) B) Split the tables, so that all non- key attributes depend on a primary key. C) Split tables further, if there are transitive functional dependencies. This results in tables with a single, primary key per table.