Alışveriş merkezleri için çeşitli çevresel faktörlere göre müşteri adedi tahminleme çalışması
Estimation of customer numbers for shopping centers according to various environmental factors
- Tez No: 609095
- Danışmanlar: DOÇ. DR. SEZİ ÇEVİK ONAR
- Tez Türü: Yüksek Lisans
- Konular: Endüstri ve Endüstri Mühendisliği, Industrial and Industrial Engineering
- Anahtar Kelimeler: Belirtilmemiş.
- Yıl: 2019
- Dil: Türkçe
- Üniversite: İstanbul Teknik Üniversitesi
- Enstitü: Fen Bilimleri Enstitüsü
- Ana Bilim Dalı: Endüstri Mühendisliği Ana Bilim Dalı
- Bilim Dalı: Endüstri Mühendisliği Bilim Dalı
- Sayfa Sayısı: 109
Özet
Veri üretiminin yıldan yıla giderek hızla artması ile birlikte veri kavramı hem dünyada hem de Türkiye'de büyük önem kazanmaktadır. Verinin önem kazanması ile birlikte veri madenciliği de değişmekte ve geliştmektedir. Veri madenciliği sayesinde firmalar müşteri yönetim stratejilerini verilere dayanarak kuracağı modeller ile belirlemeye başlamışlardır. Yapılan müşteri yönetim modelleri müşteri kazanım modelinden churn modellerine kadar çeşitlenmektedir. Bu alanda yapılan literatür taraması göstermektedir ki müşteri yönetim alanında bir çok veri modelleri ile çalışma yapılmıştır. Daha detaylı bir literatür araması yapıldığında ise, talep tahmini ve lokasyon analizinin birlikte uygulandığı ve tahmin modeli ile güçlendirildiği kaynak sayısının çok az olduğu ortaya çıkmıştır. Sektör bazlı literatür taraması yapıldığında ise alışveriş merkezleri için bu kapsamda çalışmaların çok düşük adetlerde bulunduğu gözlemlenmektedir. Bu çalışma kapsamında da literatürde bulunan bu eksikliği gidermek amacıyla alışveriş merkezleri için lokasyon analizi ve müşteri sayısını tahminleyecek talep tahmin modellerini birleştirerek yeni bir model ortaya çıkarılmıştır. Çıkarılan bu model tahminleme algoritmaları ile güçlendirilerek tüm alışveriş merkezlerine bu modelin genelleştirilebileceği test edilmiştir. Kurulan modellerde birden çok teknikler test edilmiş ve en başarılı bulunan tahminleme tekniği modelde uygulanmıştır. Bu çalışma ile alışveriş merkezlerinin sıcaklık, yağış piyasa değişkenleri, trafik yoğunluğu gibi çevre etkenleri alışveriş merkezlerine giden müşteri sayısında anlamlı bir etkisi bulunduğu ortaya çıkmıştır. Bu çalışma sonucu çıkan çıktılar doğrultusunda, alışveriş merkezleri çevre değişkenliklerini kullanarak kuracağı modeller ile pazarlama stratejilerini planlamakta ve maliyet optimizasyonu sağlayacak aksiyonlar alabilecektir. Bu kapsamda yapılan araştırmalar ve kurulan modeller detaylı olarak bu çalışmada bahsedilmektedir.
Özet (Çeviri)
With help of the developing technology for years, the world of informatics is also developed. With the effect of this situation, the data produced in one hour has expanded from gigabytes to zettabytes. However, with the growth of data, it is costly to store, store and process the data with the right methods. However, despite these costs, if the data can be stored and processed correctly, it can provide additional benefits for each sector and each business line and create new profitable business opportunities. Since the companies have also given importance to customer satisfaction in the recent periods, instead of increasing the product prices, they give importance to cost optimization and attracting new customers. Data mining and data processing are also gaining importance in taking these actions and designing their strategic plans correctly. Data mining is evolving in many sectors, products and services are emerging to meet the needs of sectors and organizations. With the spread of these services and the development of data processing technology, data mining, big data and business problems that the data will solve have become increasingly important. With the help of data mining and big data, companies started to determine their customer management strategies based on data models. Customer management strategies vary from customer acquisition model to churn models. The literature review in this field shows that many data models have been studied in the field of customer management. In recent years, with the spread of machine learning algorithms, these studies are transformed into models in which customer management strategies are determined using machine learning techniques. When a more detailed literature search was made with machine learning, it was found that the number of resources that demand forecasting and location analysis were applied together and strengthened by machine learning algorithms were very few. Within the scope of this study, a new model has been developed by combining location analysis and demand forecasting models that will estimate the number of customers for shopping centers in order to overcome this deficiency in the literature. This model was strengthened with machine learning algorithms and tested to generalize this model to all shopping centers. In the established models, multiple techniques were tested and the most successful machine learning technique was applied in the model. In this study, environmental factors such as temperature, financial market variables, traffic density of shopping centers have a significant effect on the number of customers going to shopping centers. In line with the outputs of this study, shopping centers will be able to plan their marketing strategies with models to be built using environmental variability and take actions that will enable cost optimization. As a result of seven linear regression models prepared within the scope of predictive analysis, the VIF ratio is very high if at least two of the financial data and at least two of the temperature data are included in the same analysis. This means that the linearity between variables is high. As a result of this analysis, it was found that the most appropriate method was to obtain maximum one from the financial data and maximum one from the temperature data for the correct model. In this context, the first and second regression analyzes were the most effective and successful models both in terms of test of statistical assumptions, in terms of R-adjusted and Ftest, and in terms of success rate. In addition, the first and second regression models, the normalized state of the error values when looking at the error value of the minimum and acceptable level is acceptable.This showed us that 5 independent variables had a significant contribution in estimating the number of customers. Shopping Mall Internet Trend (E_Trends), Daily Night Temperature (Night), Monthly CPI Change (CPI), Daily Average Traffic Density (Ortrafic) and E-commerce Internet Trend (E-Commerce) When the first two analyzes were compared between them, there was no E-commerce Internet Trend among the variables for the best success rate . In order to estimate the number of customers, it was observed that the best model was used as the independent variable [Shopping Mall Internet Trend (E_Trends), Daily Night Temperature (Night), Monthly CPI Change (CPI) and Daily Average Traffic Density (Ortrafic)]. In order to test the validity of the first and second regression analysis based on the machine learning algorithm established for the selected shopping center in other shopping centers and general spreading situation, these models were applied in two new shopping centers selected after the first shopping center. The first and second regression models for the first shopping center were also applied to the second and third shopping centers, which were collected as new data. Prior to this analysis, internet trend data were collected for the second and third shopping centers, and other outsourcing data were taken from the data set at the first shopping center. Before the analysis, regression analysis hypothesis tests were performed for both the second and third shopping centers. A new data table was created by deleting the external data obtained in these hypothesis tests and analyzes were made on this data table. When the first and second regression analyzes of the second and third shopping centers were examined, it was observed that the model was significant with both F test and the R-adjusted was highly explanable. Furthermore, when Durbin-watson and VIF values were examined in the table below, these values were found to be suitable for regression analysis. When the success rates are considered in line with these results, it is observed that both models achieve at least 90% success rate for both shopping malls and this rate indicates a high success rate. In addition, when the error values are examined, it is observed that the error values are at minimum and acceptable levels. However, the success rate of the two models in the first shopping center is higher than the other two shopping centers and the error values are lower than the other two shopping centers. This situation stems from the fact that the model established was established according to the first shopping center and was specially designed for this shopping center. However, both the success of the regression tests and the high estimation rates show us that the first and second regression models can be widely spread. This shows us that the same 5 independent variables have a significant contribution in the estimation of the number of customers for all shopping centers in Istanbul. When the first and second regression models are examined, it is seen that the first regression model predicts better than the second regression model. In order to estimate the number of customers, the best model as the independent variable [Shopping Mall Internet Trend (E_Trends), Daily Night Temperature (Night), Monthly CPI Change (CPI) and Daily Average Traffic Density (Ortrafic)] Based on these results, shopping centers can estimate the number of customers for the next day with a high success rate by looking at environmental variables. In order to put these environmental variables into their models, the weather information of Istanbul and its environment, the traffic density information, daily / weekly / monthly change information of the financial market, daily change information of the internet trend of the shopping mall & the competitor of the e-commerce companies need to be gathered. Shopping centers will be able to provide both cost optimization and marketing strategies by planning the next day with these models. The shopping centers will be able to plan the number of security guards by estimating the number of customers coming, the plans of the number of stores and advisors in the shopping center will be able to draw the plans and increase the number of employees by increasing the number of employees during these hours by increasing the number of employees. In addition, it will be able to plan and optimize energy costs and other costs according to these numbers. From the point of view of marketing strategies, it will be possible to create a good experience for new customers by estimating the number of future people by making special marketing studies for them, and also to ensure that customers remain loyal with existing marketing campaigns. When shopping centers want to organize events, they can increase their occupancy rate and thus their income by using these customer number estimates. Shopping centers will also have the chance to change their strategic plans according to market situation variability or traffic density variability. This analysis was performed on a 217-day data set and the results were based on these data. However, the validity of this analysis may be reduced if the climate changes radically, financial variables change in an unusual situation and change upwards or downwards with high differences, and if the internet trend is collected in a different format. In the face of this situation, models and data sets that will strengthen these variables can be prepared in the following analyzes and the established machine learning algorithm can be further strengthened.
Benzer Tezler
- Ankara kentindeki alışveriş merkezlerinin yer seçim tercihleri ve mekansal etkileri
Location choice of shopping malls and their spatial impacts in Ankara city
ÖZLEM ERDOĞAN
- Ortak yaşam alanlarından izole edilen stafilokok suşlarının, antibiyotiklere duyarlılıkları ve virulans faktörleri
Antibiotic susceptibility and virulence factors of staphylococcus strains isolated from communal areas
BAHAR KOCA
Yüksek Lisans
Türkçe
2016
MikrobiyolojiMarmara ÜniversitesiFarmasötik Mikrobiyoloji Ana Bilim Dalı
YRD. DOÇ. DR. ERKAN RAYAMAN
- Green supply chain management for construction waste: Case study for Turkey
İnşaat atıkları için yeşil tedarik zinciri yönetimi: Türkiye uygulaması
TUĞÇE BELDEK
Yüksek Lisans
İngilizce
2015
İşletmeİstanbul Teknik Üniversitesiİşletme Mühendisliği Ana Bilim Dalı
DOÇ. DR. HATİCE AKDAĞ
- Alışveriş (çarşı) yapılarının tarihsel süreç içerisindeki gelişimi üzerine bir inceleme; Çankırı örneği
Aresearch on the development of shopping (mall) buildings in the historical process; the sample of Çankırı
SELİM KARTAL
Yüksek Lisans
Türkçe
2013
MimarlıkKarabük ÜniversitesiMimarlık Ana Bilim Dalı
YRD. DOÇ. DR. YÜKSEL TURCAN
- Female consumers' sensory expectations of shopping malls: A semantic network analysis
Kadın tüketicilerin alışveriş merkezlerinden duyusal beklentileri: Anlamsal ağ analizi
TOLGA TUVAY
Yüksek Lisans
İngilizce
2021
İletişim Bilimleriİzmir Ekonomi ÜniversitesiPazarlama İletişimi ve Halkla İlişkiler Bilim Dalı
PROF. DR. EBRU UZUNOĞLU