Download presentation
Presentation is loading. Please wait.
Published byKenneth Barton Modified over 9 years ago
1
Use of cluster analysis in exploring economic indicator differences among municipalities in Latvia Ieva Braukša University of Latvia 11.11.2011 This work has been supported by the European Social Fund within the project «Support for Doctoral Studies at University of Latvia»
2
Plan of the presentation Problem and data description Disjoint clustering results Hierachial clustering results Dendrogram analysis Conclusions 2
3
Description of situation 2009 Administrative territorial reform 119 municipalities (110 districts + 9 cities) Wide debates about borders and differences of these municipalities. Cluster analysis – possiblity to look at municipality differences another perspective. 3 DO MUNICIPALITIES GROUP BY PLANING REGIONS OR BY OTHER ASPECTS OF SIMILARITY?
4
Data used Data from State Regional Development Agency (VRAA) Variables used during clustering: – Changes in number of permanet residents (2006-2011) – Share of residents at working age (1.1.2011) – Level of unemployment (1.1.2011) These variables include basic information about inhabitant structure and economic conditions. Data are standartized (because variables have different measurement units, so standartized to avoid influence of different variable variance) – mean 0, std 1. 4
5
Methods of analysis Disjoint clustering Hierarchial clustering + dendrogram 5
6
Disjoint clustering 6
7
Results of disjoint clustering * ClusterInhabitant change Working age Unemployed 1 -0.59-0.481.74 (0.25)(0.791)(0.954) 2 -0.080.77-0.27 (0.474)(0.554)(0.397) 3 3.000.90-1.17 (1.158)(0.872)(0.164) 4 -0.24-0.80-0.31 (0.42)(0.682)(0.497) 7 *using FASTCLUS clustering method in SAS® software Table showing cluster means and standard deviations
8
Cluster group diferences 8
9
9
10
10
11
Cluster 1 Less than average share of inhabitants at workig age Highest unemployment Fastest decrease of permanent residents 11 Latgale 16 Vidzeme 1 Zemgale 1 Kurzeme 1 Aglonas novads Aluksnes novads Auces novads Baltinavas novads Balvu novads Ciblas novads Dagdas novads Karsavas novads Kraslavas novads Livanu novads Ludzas novads Priekules novads Rezeknes novads Riebinu novads Rugaju novads Varkavas novads Vilakas novads Vilanu novads Zilupes novads Municipalities in cluster:
12
Cluster 2 Relatively larger share of inhabitants at working age “Mainstream” 12 Kurzeme 4 Latgale 2 Pierīga 12 Vidzeme 12 Zemgale 13 Aizkraukles novads Aknistes novads Alsungas novads Bauskas novads Beverinas novads Burtnieku novads Cesu novads Daugavpils novads Dobeles novads Gulbenes novads Iecavas novads Incukalna novads Jaunjelgavas novads Jaunpils novads Jelgavas novads Keguma novads Kocenu novads Kokneses novads Krimuldas novads Lielvardes novads Madonas novads Malpils novads Nauksenu novads Neretas novads Ogres novads Olaines novads Municipalities in cluster: Pargaujas novads Preilu novads Priekulu novads Raunas novads Ropazu novads Rundales novads Salas novads Salaspils novads Saldus novads Sejas novads Siguldas novads Smiltenes novads Talsu novads Tervetes novads Vecpiebalgas novads Vecumnieku novads Ventspils novads
13
Cluster 3 The only cluster with significant increase of number of permanent residents Largest share of inhabitants at working age Smallest levels of unemployment 13 Pierīga 8 Adazu novads Babites novads Carnikavas novads Garkalnes novads Ikskiles novads Kekavas novads Marupes novads Stopinu novads Municipalities in cluster:
14
Cluster 4 The smallest number of customers at working age Not high unemployment, decrease of permanet residets – moderate 14 Kurzeme 12 Latgale 1 Pierīga 8 Vidzeme 12 Zemgale 6 Aizputes novads Alojas novads Amatas novads Apes novads Baldones novads Brocenu novads Cesvaines novads Dundagas novads Durbes novads Engures novads Erglu novads Grobinas novads Ilukstes novads Jaunpiebalgas novads Jekabpils novads Kandavas novads Krustpils novads Kuldigas novads Ligatnes novads Limbazu novads Lubanas novads Mazsalacas novads Nicas novads Ozolnieku novads Pavilostas novads Plavinu novads Municipalities in cluster: Rojas novads Rucavas novads Rujienas novads Salacgrivas novads Saulkrastu novads Skriveru novads Skrundas novads Strencu novads Tukuma novads Vainodes novads Valkas novads Varaklanu novads Viesites novads
15
Cluster summary Group 1 Unemployment Decrease of number of inhabitants Group 2 More inhabitants at working age Group 3 More employed Increase number of ihabitants Larger share of working age population Group 4 Less inhabitants at working age 15
16
Hierarchial clustering 16
17
Hierarchial clustering methods Several used to test if results are similar: – average linkage (group average, unweighted pair- group method using arithmetic averages) – centroid method (unweighted pair-group method using centroids, centroid sorting, weighted-group method) – complete linkage (furthest neighbor, maximum method) 17
18
Average linkage method 18
19
Centroid method 19
20
Compete linkage method 20
21
21 Average linkage method – dividing dendrogram in groups
22
Group 1 – municipalities from Latgale 22 Note: P – Pierīga, L – Latgale, V – Vidzeme, Z – Zemgale, K - Kurzeme
23
23 Group 2 – mainly Vidzeme Note: P – Pierīga, L – Latgale, V – Vidzeme, Z – Zemgale, K - Kurzeme
24
24 Note: P – Pierīga, L – Latgale, V – Vidzeme, Z – Zemgale, K - Kurzeme Group 3
25
Group 4 25 Note: P – Pierīga, L – Latgale, V – Vidzeme, Z – Zemgale, K - Kurzeme
26
26 Group 5 Note: P – Pierīga, L – Latgale, V – Vidzeme, Z – Zemgale, K - Kurzeme
27
Group 6 - Pierīga 27 Note: P – Pierīga, L – Latgale, V – Vidzeme, Z – Zemgale, K - Kurzeme
28
Conclusions Cluster analysis based on inhabitant structure and basic economic indicator analysis shows that : – There are some regional similarities – dendrogram shows two distinct groups for Pierīga and some Latgale municipalities; – Other planing regions don’t create separate groups. There are similar municipalities across all of them. 28
29
Thank you! Ieva.Brauksa@inbox.lv
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.