Yüksek Lisans, Proje Koleksiyonu

Permanent URI for this collectionhttps://hdl.handle.net/20.500.11779/215

Browse

Now showing 1 - 20 of 112

A Case Study on Churn Prediction and Understanding Customer Behavior
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2018) Kıralı, Gülşen; Tönük, Gökçe
Churn prediction is essential for businesses as it helps to detect customers who have the potential to cancel a subscription to a product or service. Churn prediction techniques try to understand the certain customer behaviors and attributes which signal the risk andtiming of customer churn. Companies started to focus on retention activities more in the last years since holding current customer in the system is less costly when compared with acquiring new ones. In order to allocate costs to right customers, companies prefer to use the big part of these budget to potential churn customers which makes the accuracy of churn detection important for us. The objective of this project is to develop a machine learning algorithm that predicts potential churn customers that will not make any transactions in the following three months. While predicting churn, some customer segments and subsegments are created in order to understand the common behavior of potential churn customers. Common characteristics of loyal customers will also beinvestigated in order to determine churn prevention marketing activities for potential churn customers. Among all of the machine learning algorithm trials including Logistic Regression, Boosted Decision Tree, Support Vector Machines, Decision Forest, Decision Forest Regression and Neural Networks, Logistic Regression predicts with the highest accuracy and lowest number of False Negative which means model slightly mistaken unchurned customers.
A Comprasion of Ensemble Learning Methods in Retail Sales Forecasting
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2019) Süer, Serhan; Güney, Evren
Forecasting has always been an essential skill which companies try to have and implement in various areas. Sales forecasting is one of the major usage areas of forecasting which is used in almost all sectors. This study refers to forecasting sales of Walmart Stores based on several features such as store id, department id, date, and store size. Walmart sales data which was used in this study contains information of stores between 2010 and 2012. At the beginning of the study, the introduction of the dataset and exploratory data analysis were made to identify dependent/independent variables and their characteristics. To apply machine learning algorithms, data preprocessing methods such as missing value treatment, outlier treatment, and feature selection was applied. Ensemble learning methods in machine learning algorithms were applied in the modeling stage. These methods were addressed in three parts such as Bootstrap Aggregation, Boosting, and Stacked Generalization and these parts consist of six different algorithms in total. The models were compared based on four regression metrics as Root Mean Square Error, Mean Absolute Error, R-Squared, and runtime. After selecting the main metric which models were evaluated, cross-validation was applied to achieve unbiased estimates. Finally, parameters of the model which have the highest score in cross-validation were tuned in the hyperparameter optimization stage and a machine learning model which can be used in forecasting sales of Walmart stores and its success score were obtained.
A Study on Churn Prediction in Telecommunication and Pay Tv Area
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2020) Şayık, Murat; …
...
Ad Click Prediction Using Machine Learning Algorithms
(MEF Üniversitesi Fen Bilimleri Enstitüsü, 2021) Uncu, Nazlı Tuğçe; Hande Küçükaydın
Online advertising has a great potential to boost business’ revenue. One of the key metrics that defines the success of online ad campaigns is click through rate (CTR) which indicates the total number of clicks received in relation to the total impression. Therefore, the click prediction systems, which have the aim of increasing the click through rates of online advertising campaigns by predicting the clicks, have become essential for businesses. For this reason, predicting whether an advertisement will receive a click fromthe user or not attracts the attention of researchers from the both industry and academia. In this capstone project, the click prediction is studied by using Avazu’s click logs dataset. The effects of having high cardinality categorical features and imbalanced data are examined during data preprocessing phase and then relevant features are selected to be used in modeling. The methods that are used for this classification problem are decision trees, random forest, k-nearest neighbor, extreme gradient boosting, and logistic regression. According to the results of the study, extreme gradient boosting shows the best performance.
Airbnb Host Recommendation Engine
(MEF Üniversitesi Fen Bilimleri Enstitüsü, 2021) Arslan, Batuhan; Özgür Özlük
In this project, a fifth rule is proposed to reveal guests ' comments about hosts using the recommendation system and sentiment analysis for the super hosts' selection for Airbnb. This project is aimed to contribute to Airbnb's selection of Super hosts. In this study, sentiment analysis and comment data are examined, and polarity scores are created for use in suggestion systems. A collaborative filtering method is used for the recommendation system. The FunkSVD algorithm received the best RMSE score. Polarity scores are estimated for each latent user by looking at the host and listing id. The recommendation system developed ranked the polarity scores of hosts for each user.
Alternative Credit Scoring Model for Thin File Customers
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2019) Korkmaz, İstem Akça; Taş Küten, Duygu
Credit scoring is a widely used tool for banks, financial institutions or corporations. Traditional credit score models are calculated from past financial history of users, and this may lead to exclude some people who have limited financial history from the credit system. Alternative credit scoring allows sector players to access to a larger portion of these customers. The credit scoring industry has expanded with an "all data is credit data" approach that combines traditional credit scoring systems with new data points. In this study, we aim to build an alternative credit scoring model for customers who have limited financial historical data (thin file) by using alternative data points for a national bank in Turkey. Some of the alternative data points and variables have been gathered from one of the bank’s products: the authorized card for Turkish national league football tickets (Passolig). Using alternative data points combining with demographical and geographical information, we perform a comparison between the machine-learning approaches. We use logistic regression approach as a base model and perform a comparison between tree-based approaches: decision tree, random forest and XGBoost to select the most effective modelling approach
An Examination of the Effect of Monetary Expansion Policies Implemented by Four Large Central Banks After the 2008 Global Financial Crisis and the Covid-19 Crisis on Developing Countries on the Example of Turkey
(MEF Üniversitesi Sosyal Bilimler Enstitüsü, 2021) Özbey, Sakine Gülşah; Nazlı Karamollaoğlu
After 1980, financial markets took a share due to globalization trends in the world. In literature, many studies exist which show us that the financial crisis and financial globalizations started to appear more often than it did in the past. The market’s mood is reflected in the data when risks and incalculability increase in financial markets. Financial liberalization and the removal or significant reduction of inspections have increased the fragility of markets. A number of decisions were made and interfered with by many authorities after the financial crisis which was felt all around the world for a long time. Before the 2008 economic crisis price stability was a focus for central banks. However, the importance of financial stability came into prominence after thecrisis. The negative effects of Covid-19 crisis, which was not originated from economic reasons at the same time, which created a supply and demand shock, were seen fast. Like in every crisis politicians interfered in order to reduce the effects of the crisis. The connection between the 2008 global financial crisis and Covid-19 crisis is the need to increase declining total demand. By the reason of reduced economic activity on a global scale, monetary and fiscal policies and inventions that increase economic activity have been involved. The concept of globalizations has multifaceted effects ondeveloping countries. By the entering of funds into enhanced market economies, it helps developing countries to meet the need for financing that will provide economic growth and development, while reducing production and increasing dependence on external financing. With financial globalization direction and momentum of the movement of fund is changing according to countries’ macroeconomic appearance. Particularly development and decisions taken in countries like the United States and England, which have the right comment on world trade, have influenced all around the world. The policies implemented by these countries in times of crisis are closely followed by economic actors. In this study FED (Federal Reserve Bank), ECB (European Central Bank), BOJ (Bank of Japan), BOE (Bank of England), monetary easing policies implemented by central bank after the global crisis in 2008 and the Covid-19 crisis were examined and how developing countries are affected by these crisis and policies are discussed and the example of Turkey was examined.
Analyzing the Drivers of Customer Satisfaction Via Social Media
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2019) Yücel, Kadir Kutlu; Koç, Utku
Social media became a great influence force during the last decade. Active social media user population increased with the new generations. Thus, data started to accumulate in tremendous amounts. Data accumulated through social media offers an opportunity to reach valuable insights and support business decisions. The aim of this project is to understand the drivers of customer satisfaction by public sentiments on Twitter towards a financial institution. Data was extracted from the most popular microblogging platform Twitter and sentiment analysis was performed. The unstructured data was classified by their sentiments with a lexicon-based model and a machine learning based model. The outcome of this study showed machine learning based model successfully overcame the language specific problems and was able to make better predictions where lexicon-based model struggled. Further analysis was performed on the extreme daily average sentiment scores to match these days with prominent events. The results showed that the public sentiment on Twitter is driven by three main themes; complaints related to services, advertisement campaigns, and influencers’ impact.
Applications of Balanced Scorecard in Different Business Contexts
(MEF Üniversitesi Sosyal Bilimler Enstitüsü, 2019) Aktaş, Utku; Karadağ, Hande
In this project, it is aimed to explain the elements and give background information, to show industrial applications of balanced scorecard in different industries, to show implementations of balanced scorecard in 4 chosen companies, to explain corporate balanced scorecard in academic studies, and last but not least conclude the research. The results of this analysis indicates that this document is a very important tool for a company to track and measure its success, detect and identify the relevant measures and actions in order to achieve the overall goals. It is seen that even the single work of an employee can highly contribute to the goals of the company, if integrated with balanced scorecard. This study is expected to contribute to the literature by its focus on the important performance measurement system of Corporate Balanced Scorecard and the application of this system in different contexts.
Association Rule Mining on Ticket Sales Data
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2018) Genç, Özge; Küçükaydın, Hande
This study aims to analyze the ticket sales data of a cultural institution and define association rules between the festivals/event group and festival/event group venues by market basket analysis. Market basket analysis is a well-known data mining method that is used to discover similarities between products or product groups. For market basket analysis, the apriori algorithm is applied which is considered as a popular data mining algorithm and helps to discover frequent item sets and forms association rules within the dataset. In this project, the apriori algorithm is applied using Python to discover the association rule: between the venues (implementation 1), between the venues excluding the venues used for a specific festival type (implementation 2), between festivals and event groups if any (implementation 3). According to the results of implementation 1, the associations are mostly between the venues of a specific festival type. According to the implementation 2, when we exclude this specific festival type from the dataset, it is seen the rules are mostly between the venues of another festival type. In implementation 3, when the festival venues variable is excluded and only the event names are considered, it is seen that the support, lift and confidence values are lower than the previous implementations.
Benchmarking of Recommendation Models for an On-Line Fast Fashion Retailer
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2018) Tilkat, Mustafa; Küçükaydın, Hande
This project studies the usage of the recommendation engines to improve the sales in an online fashion retailer. Fashion retailers sale variety of products throughout their online channels. Since the number of products can be huge compared to an in-line shop, customers may miss some of them while shopping online. Hence, it is crucial to display products that are more likely to be purchased by a customer when the customer is surfing on the website. Our problem is motivated by practice at an online fashion retailer in Turkey. Four collaborative filtering-based algorithms and a random recommender are utilized to design a recommendation engine. 80% of the data is used for training while the other 20% is to used test the designed method. Based on our experiments, User Based Collaborative Filtering (UBCF) using Pearson correlation outperform the other algorithms based on Receiver Operating Characteristic (ROC) curve.
Big Data Analysis on Hotel Reviews
(MEF Üniversitesi Fen Bilimleri Enstitüsü, 2021) Demir, Burcu; Özgür Özlük
This analysis aims to get a regression model of the reviews and the score by the guests to observe the effects of the content of the reviews on scores. The content of the reviews is also suitable for a sentiment analysis. These analyses are useful indicators of the hotel sector to catch the market direction positively. In this analysis, clustering hotel-based reviews and customer segmentation based on the reviews will be the key point. Nationality of the guests will be helpful information of the guests to get them into the segmentation pool. The guest who wants to stay in the best hotel in Europe while their trip could choose the best hotel. They can conclude that selection by meeting their needs.
Big Data Analytics on Used Car Information
(MEF Üniversitesi Fen Bilimleri Enstitüsü, 2021) Demir, Efe; Utku Koç
In this research, a decision support system is implemented on a used car dataset. The main purpose is to predict the price information and reveal the related features. The price prediction problem is classified as a regression problem. The key point is to find the best-fitting model and obtain the best accurate prediction outcomes. Should we buy this car, or at what price may I sell my car? This work is about to answer these questions. Various regression models are compared, and detailed results are explained correspondingly.The constructed models will help customers to know about their car price and salability. And they can identify the buying opportunities. The percentage error approach which is detailed in the results section will be a guideline for customers/firms to make a market analysis or detect fraudulent listing information.
Boarding Pass Detection In Social Media To Prevent Flight Information Thft
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2019) Ekici, Hasan Oktay; Çakar, Tuna
During the past few years, along with social media gaining bigger share on people’s lives, everyone has started to share their moments with detailed information on multiple platforms instantly. Sharing these kinds of information on posts may cause security bugs in people’s lives, such as undesired flight changes/cancellations because of flight information theft, frequent flyers miles theft, and even car theft and burglary. This project’s aim is to develop an artificial intelligence algorithm that can help to prevent these security bugs. In this project, we use data that is collected from instagram posts that contain boarding passes. Our main purpose is to build an artificial intelligence that makes decisions and processes following procedures: a machine learning algorithm that decides if the shared instagram post contains a boarding pass shared with #boardingpass; an optical character recognition algorithm that gathers text information from the post and scripts that send the information instantly to the relevant air carrier about the shared post. With this information, air carrier will be able to inform the passenger about their concern on the flight safety only in a couple of minutes after the post is shared.
Building Footprint Extraction Using Deep Learning Techniques
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2018) Deniz, Oytun; Gökmen, Muhittin
Geospatial industry is getting bigger and bigger these days in addition to creating massive amount of data. Today map features such as roads, building footprints are created through manual techniques. There is not automated solution that extracts map features such as roads, building footprints from satellite imagery. Advance automated feature extraction techniques will serve important uses of map data including disaster response. SpaceNet is a commercial satellite imagery and labeled training data to foster innovation in the development of computer vision algorithms. In this paper we will give a brief explanation about image classification, object recognition processes and why deep learning is effective on object recognition, and how we can apply these concepts to our problem which is Building Footprint extraction. And we will use SpaceNet’s dataset and apply tensorflow backhand object detection model.
Calculation of the Capacity of a Retail Clothing Store
(MEF Üniversitesi Fen Bilimleri Enstitüsü, 2021) Türkoğlu, Murat; Tuna Çakar
The purpose of this study is to define and improve the capacity calculation for clothing store companies. It is important to know how many products need to be sent to the relevant store in order to sell more products, whether more or less products will be sent to the stores, how much the product can sell and how much capacity the stores will have for the relevant products during the season. Planning and producing more products than necessary may also cause insufficient capacity to consume the stocks of that product in the relevant season. For these reasons, a detailed capacity management system is needed. The capacity of certain product groups in the stores in certain seasons can be determined by calculating the capacities of the relevant units in the stores and the relations of these units with the product groups. The relevant system will produce output for both planning and allocation units. At the same time, converting the capacity of products and store display units into a common unit (LCM) will be one of the factors that facilitate our work in capacity calculation. A short version of the LC Waikiki capacity system platform is used to obtain the data. ASP.Net Core Web API, ADO.NET, T-SQL, C # programming were used as program tools. Azure Microsoft SQL Server was used as a database server. Azure App Services has been used to keep the business codes.
Carbon Price Forecasting
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2018) Karakaya, Nurhak; Ağralı, Semra
In last twenty years great improvements occurred both in technological advances and in the world economic capacity. The total production capacity of countries has been increasing rapidly. These increases need great usage of energy. For that reason, prices of energy related products are very important as they dramatically affect company budgets. Energy budgets get a great deal in total budget of companies and countries. A unit increase in an energy related product can severely affect the budget. The carbon price is one of those products. Besides carbon prices, carbon usage also affects global environment so its price also has an impact on global temperature. To forecast future carbon price different machine learning methods are used. In literature, support vector machines (SVM) [1, 2, 3], random forest (RF) [4, 5], artificial neural networks (ANN) [6, 7, 8] and Auto Regressive Moving Average (ARMA) [9] are commonly used methods. All these methods have pros and cons over the others. In this project, we also apply different machine learning methods, ANN, SVM, RF, Lasso Regression (LG)[11] and Ridge Regression (RR) [10] to forecast the carbon price over time, and give an explanation for future price movements. Then, we compare those five models by analyzing model validation methods. Finally, we choose the best model for further experiments. We have four data types: daily carbon price (CP), electricity price (EP), natural gas price (NG) and coal price (COP) that cover the period of 2009 and 2017. Prices are provided in different currencies. First of all, we work on the data to have all prices in the same currency. We completely eliminate null data. Then, graphically we investigate overall trend by smoothing the data. For analyzing data, we look for daily, monthly, yearly and seasonally time scales. For every weekday or weekends in train data set we keep a day in test data set so that we can keep the time effect in our model. After the data management process, we apply different forecasting methods to explain future carbon price tendencies.
Chuen Analysis of Gittigidiyor Customers
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2017) Kantarcı, Özlem Hazal; Çakar, Tuna
In this project, it is aimed to estimate the loyalty of the customers of the e-commerce company named GittiGidiyor by analzying the customer movements and examined which movements affected the customer loyalty positively / negtively. In the dataset studied, it was seen that the number of active customers is much higher than that of passive customers. Several methods have been tried to solve this "Class imbalance" problem and it has been decided to replicate some lines of passive customers. Rows of smaller classes are duplicated to compensate classes with generated code. The data set was divided into training, validation and test and different algorithms were used. One of the innovative approaches was training and validating models in an earlier time window and testing the model with samples from a later time window. As a result of the studies, it was decided to use "Linear Discriminant Analysis" considering its short training time and especially the success of predicting passive customers.
Churn Prediction in Vodafone Turkey
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2017) Genel, Gökhan; Ağralı, Semra
This Capstone Project focuses on finding a churn pattern in Vodafone postpaid consumer subscribers. The churn value refers to disconnection from subscription via port-out / Mobile number portability (MNP). It is one of the most important items that demonstrate revenue-loss. The subscriber who churned with MNP switches to a rival GSM operator. The cost of keeping an existing customer is generally cheaper than the cost of acquisition of a new customer. Focusing on customer retention is one of the most profitable strategy for growth. Statistical analysis and machine learning can help analyze churn activities and they can even alert companies when their existing customers are likely to churn. By using machine-learning algorithms, this project aims to detect Vodafone postpaid consumer subscribers who are likely to churn. This project will help the company to decrease its revenue loss.
Churn Prediction of a Deal E-Commerce Website Customers
(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2017) Çevik, Müge; Küçükaydın, Hande
Today, there is a lot of deal e-commerce sites which are essentially marketplaces. They provide deals which are offered by merchandisers. Because of the nature of these sites there is no subscription model; customers continue because of price or interest or quality not because of subscription. It is normal to have some customers who stop buying, which is defined by "churn". Data mining is now a new technique to define "churned" customers and to have prediction who will churn and what should be against. In this project customers are clustered via unsupervised clustering technique for clusters as "newly purchased", "frequently purchased" and "mostly payed" and "churned". Random Forest Classifier is used to prove that the "churned" customer clusters have homogeneous character and also it has been proved that the "churned" labelled customers have actually no deal order after the observed time period. To recommend what should be done to regain the churned customers to the site the deal order history of these customers have been explored and the deal categories from which they have bought have been found.

Browse

Browsing Yüksek Lisans, Proje Koleksiyonu by Language "en"