Ekonomik Faaliyet Kollarında COVID-19 Pandemi Etkisinin Çok Boyutlu Ölçekleme ve K-Ortalamalar Kümeleme Analiziyle İncelenmesiMuhammet Atalay
COVID-19 pandemisinin tüm dünyaya yayılmasıyla ekonomik faaliyetlerde küresel bazda önemli değişiklikler meydana gelmiştir. Türkiye bu durumdan global ve lokal bazda önemli düzeyde etkilenen ülkelerdendir. Çeşitli ekonomik faaliyet kollarında iş yeri ve istihdam sayıları bu etkinin gözlenebildiği önemli göstergelerdendir. Bu çalışmada; pandeminin hemen öncesi (2019 yılı) ile hızlı ve yoğun olarak görüldüğü erken dönem (2020 yılı) iş yeri sayıları ve zorunlu sigortalı çalışan sayıları, iller bazında incelenerek faaliyet kollarına göre pandeminin etkisinin ortaya çıkarılması amaçlanmıştır. Yöntem olarak istatistiksel veri analizi ve veri madenciliği tekniklerinden çok boyutlu ölçekleme ve kümeleme analizleri kullanılmıştır. Bu yöntemler yardımıyla elde edilen bulgular görselleştirilmiş ve çalışmanın amacı doğrultusunda yorumlanmıştır. Elde edilen sonuçlara göre, iki yılın verileri karşılaştırıldığında, toplamda iş yeri sayısı ve zorunlu sigortalı çalışan sayısının arttığı görülmüştür. Faaliyet kolları bazında sonuçlar incelendiğinde değişimlerdeki pandemi etkisi göze çarpmaktadır. Mobiliteye dayalı ve pandemi tedbirlerinin engellediği faaliyet alanlarının iş yeri ve çalışan sayısı bakımından azalma yönünde etkilendiği görülmüştür. Öte yandan bu kısıtlamaların özellikle perakendecilik sektörlerini dijital ortamlara taşıyarak e-ticarette büyümeye sebep olması, posta ve kargo faaliyetlerinde yüksek oranlı artışa neden olmuştur. Bunun yanı sıra evde bakım faaliyetlerinin de pandemi etkisiyle en fazla artışın olduğu kollardan olduğu sonucuna ulaşılmıştır.
Investigating the Impact of the COVID-19 Pandemic on Economic Activities Using Multidimensional Scaling and K-Means Clustering AnalysisMuhammet Atalay
With the spread of the COVID-19 pandemic all over the world, significant changes have occurred globally with regard to economic activities. Turkey is one of the countries to be affected by this situation on a global and local basis. The number of workplaces and employment in various segments of economic activity are important indicators through which this impact can be observed. These changes have occurred locally in different regions and different lines of business. This study aims to reveal the pandemic’s impact by examining by province the number of workplaces and number of employees with compulsory insurance just before the pandemic (2019) and in the pandemic’s early period in 2020 when it was seen spread rapidly and intensely. The study uses multidimensional scaling and clustering analyses from the statistical data analysis and data mining techniques as the research methods. The findings obtained with these methods have been visualized and interpreted in line with the purpose of the study. When comparing the data of these two years in accordance with the obtained results, the number of workplaces and the number of employees with compulsory insurance were seen to have increased overall. When examining the results on the basis of operating segments, the pandemic is seen to have had a striking impact with regard to the changes, with the operation segments based on mobility and on those prohibited by the pandemic measures being observed to have been affected by a decrease in terms of the numbers of workplaces and employees. Meanwhile, these restrictions led to growth in e-commerce, particularly by moving retail sectors to digital environments, and this caused a high rate of increase in postal and cargo activities. Home care activities were additionally concluded to be among the segments with the highest increase due to the pandemic’s effects.
COVID-19 has affected the whole world economically in addition to human health. The measures taken have completely changed social and economic life, with many sectors having been forced to stop or slow down due to restrictions. Turkey is one of the countries whose economy has most felt the effects of the global pandemic. These effects were observed to manifest in different ways with regard to the line of business and sector. While some branches of activity have experienced expansion, others may have experienced contraction. These changes have also had global as well as local effects. When considering that the disease came to the fore in the world as of the end of 2019, examining 2020 in comparison with 2019 will allow the first effects of the crisis to be seen.
This study aims to take a different view of the changes experienced by the Turkish economy due to the impact of the COVID-19 pandemic based on the number of workplaces and the number of employees with compulsory insurance with respect to province for the years 2019 and 2020. Turkey has regions with different characteristics due to its geographical and economic structure. The economic activity structure of each province may differ, and the local effects are thought to be revealable by examining the provinces.
This study uses the statistical data analysis and data mining techniques of multidimensional scaling and clustering analyses with the support of data visualization as the research methods. The cross-industry standard process for data mining (CRISPDM) model was used in the design of the study’s method section. The steps of CRISPDM are as follows: 1) define the problem, 2) understand the data, 3) prepare the data, 4) set up the model, 5) evaluate and select the model, and 6) apply the model.
The study’s dataset consists of the variables of the number of workplaces in the economic operating segments and the number of insured workers with respect to each of Turkey’s provinces. The study used the data published by the Republic of Turkey Social Security Institution (SGK) for the years 2019 and 2020. When comparing the data from the two years, the numbers of workplaces and employees with compulsory insurance are seen to have increased overall.
The steps taken while preparing the data for analysis are as follows: Because some provinces have no workplaces regarding certain operating segments, missing data were first arranged. For this purpose, the data for the operating segments were first obtained without omitting any of these areas by collecting the information for the segments where data were available in each province. In addition, operating segments with missing data were deleted and a data set was obtained with the remaining 47 operating segments. Because the aim of the study is to examine the changes that occurred in the operating segments due to COVID-19’s impact, new variables were obtained by transforming these datasets and dividing the 2020 values with the 2019 values. The new datasets obtained with these ratios were used in the multidimensional scaling and clustering analyses.
Multidimensional scaling (MDS) is a multivariate data analysis approach used to visualize similarities and uniqueness between samples by turning units into points and plotting them on k-dimensional graphs. The MDS algorithm takes as input data a similarity or differences/dissimilarity matrix representing the distances between pairs of objects. In cases where this matrix (i.e., the distance matrix) can be obtained, metric scaling can be performed. In metric scaling, the original distances between objects and the distances on the map that have been calculated by scaling are scaled the same. This study has preferred the Euclidean Squared distance metrics.
Cluster analysis is a method that aims to group objects or variables in the data matrix according to the values they take. When clustering, the distance or similarity matrix is firstcreated using the distance and similarity measures from the data matrix. Clusters are then created using the determined clustering method. Clustering methods are divided into hierarchical and non-hierarchical. In non-hierarchical methods, the number of clusters is determined in advance, and a single optimal clustering result is obtained according to the number of clusters determined using distance or similarity measures. This study uses the k-means algorithm, which is one of these methods. The Euclidean Squared distance metric was used as the metric distance. In this way, the study aims to use the outputs from the two analyses by overlapping them.
The analyzes were interpreted through visual mappings obtained in 2-dimensional space. In addition, the findings were made more understandable by comparing them with the values in the data set. According to the analysis findings, the pandemic has had a striking impact with regard to operating segments. The numbers of workplaces and employees were observed to have decreased with regard to the operating segments that are based on mobility or that had been prohibited due to pandemic restrictions and measures. The measures that were taken were determined to have caused high increases in postal and cargo operations as well as home care operations.