Yüksek Lisans Tezleri
Permanent URI for this collectionhttps://hdl.handle.net/20.500.11779/1785
Browse
7 results
Search Results
Master Thesis Interviewster: A chatbot evaluating competency based interviews using transformer models(MEF Üniversitesi, 2022) Atıcı, Onur Emre; Demir, Şenizİşe alım, insan kaynaklarının en sözel ve iletişimsel alanlarından biridir. Bu departmanın insan faktörünün baskın olması nedeniyle yeniliğe açık olduğu kadar önyargıya da açık olan birçok yönü bulunmaktadır. Bu da yapay zeka teknolojilerindeki ilerlemeyle birlikte birçok inovasyon ihtiyacını (ve şansını) beraberinde getirmektedir. Bu çalışmada adayları karşılayan, bilgi toplayan (ad-soyad, iş durumu, bilgisayar bilgisi, eğitimi, hobileri gibi) ve geçmiş deneyimleri hakkında yetkinlik bazlı sorular sunan ve bu soruları doğru cevaplayabilmesi için onlara yardımcı olan "Interviewster" adlı bir sohbet robotru oluşturmaya odaklanılmaktadır. Bu sohbet robotu adayı karşılar ve konuşmayı başlatır, adaydan toplanan verileri kaydeder, yetkinlik bazlı görüşme yapar ve sinir ağları mimarileri ve transformer tabanlı teknolojileri kullanan doğal dil işleme teknikleri ile adayın gerekli yetkinliğe sahip olup olmadığına karar verir. Web üzerinde çalışmakta olan bu sohbet robotu Python ile kodlanmış ve Flask ile web'de yayınlanmış olup Mysql veritabanını kullanan bir Python çekirdeği üzerinde çalışmaktadır. Bu tezde ilk olarak mülakat uygulamaları tanıtılmakta ve yetkinlik bazlı mülakatların yöntem ve uygulamaları anlatılmaktadır. Sonrasında Interviewster olarak adlandırılan sohbet robotunun mimarisi, kullanılan teknolojiler, kütüphaneler ve makine öğrenmesi teknikleri, detayları verilerek açıklanmıştır. Son olarak da transformer tabanlı modeller olan BERT, DeBERTa ve ELECTRA modellerinin gerçek adayların yetkinlik bazlı mülakat sonuçlarına uygulandığı bir değerlendirme çalışmasının sonuçları detaylı olarak tartışılmıştır.Master Term Project Flight Delay Prediction(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2019) Kurt, Mustafa; Taş Küten, DuyguThis study aims to create a model to predict flight departure delays. Various factors might affect a flight delay, and thus different features might be selected as input to create a model concerning priorities and the power of control over the features for the party who makes the analysis. In this study, domestic commercial flights in the U.S. operated in August 2018 are studied. Besides, airplane, passenger boarding, and cargo data are combined with flight data to benefit from possible insights related to these factors. For predicting the flight delays, machine learning methods such as decision trees, random forest, bagging classifier, extra trees classifier, gradient boosting and xgboost classifier are used and results are analyzed. Further studies could be adding extra features such as data related to flight planning, personnel data, loading data, data about technical processes to prepare a plane to a flight to improve prediction capacity.Master Term Project Suicide Tendency Classification and Suicide Number Prediction Forpopulation Subgroups(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2019) Ak, Mehmet; Küçükaydın, HandeSuicide is becoming a bigger problem for the world day by day and detecting population subgroups who are more prone to suicide is seen as one of the most important steps for taking precautions to decrease the suicide rates. This study consists of five machine learning models for suicide tendency classification and three machine learning models for prediction of suicide numbers by population subgroups. The dataset provided by World Health Organization is used in the project. Obtained models classify population subgroups as suicide-prone or less suicide prone with 86% accuracy and explain 90 % of the variance in the suicide number per 100,000 population of specific countries.Master Term Project Retention Period Prediction for Pension Policies(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2019) Bayır, Ömer; Güney, EvrenCustomer Retention in Pension market refers to the activities and actions companies and organizations take to reduce the number of customer defections. How long the customer will be with our company or will stay in the system is retention. There are already workings in my company and other companies in the market about customer retention. Existing works generally contains how to measure customer retention and how to define distribution channels are successful in customer retention. Also existing predictive models are working on the feature set customer fund, total collection, un-paid premium frequency in general. In pension market companies have small margin of profit from pension policies. To make a profit from pension policies the companies have to retain their customer for long years. It ‘s approximately nine year to make profit from a pension policy because of high sales costs. Therefore to gain a new customer is less profitable than retaining present customers in Pension Market. In my project, I want to look retention in the pension application phase of customer. My main purpose is when the customer applied for pension product predict its retention period. If I produce an applicable model, It will be used in my company’s sales channels.Master Term Project Predicting Customer Satisfaction Via Structed and Unstructured Data Using Classification and Regression(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2019) Danışman, Efehan; Küçükaydın, HandeAccording to different studies, retaining existing customers is five or more times more costly than acquiring new ones. This study aim to understand what customers expect from an airline using machine techniques. Dataset is scraped from Skytrax’s Airline Quality website and consists of 65947 observations with 17 columns consisting of one free format column that includes customer review. In order to do predict whether a customer recommends an airline or not, we try to utilize classification and regression algorithms. In addition to insights, this study also aims to compare the performance of the models and viability of using only free text in order to predict customer satisfaction.Master Term Project Predicting Ompact of Product and User Features on the Sales in an E-Commerce Site(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2018) Boyacı, Samet; Yıldırım, İrem ZeynepIn recent years, the ratio of online shopping to total shopping has been increasing continuously. Many factors affect sales of e-commerce sites. Prior to purchasing, users are concerned whether the features of the products they are interested in match their own needs or not. In this study, the most important factors in the sales of the products which are the features of the products, attributes of the sellers and interactions with product investigated. A model was developed based on available fashion products in a market where individual users could sell second-hand textiles and accessories. Using this model, we tried to predict which products would be sold by examining at the features of the products, attributes of sellers and interactions with the product. Different algorithms were investigated for predicting sales and the results were reported. By comparing the outputs, the most successful algorithm and the most important features affecting sales were identified. As a result of this study, it was determined that the most efficient algorithm was the decision tree model. When the inputs of the model were examined, it was determined that the most important features affecting salability were the interactions with the products such as the number of likes and bids.Master Term Project Underlying the Bias for Human Music Evaluation(MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2017) Yıldırım, Burak; Çakar, TunaPredictive analysis is the process of using data analytics to predict the future over historical data. Data analytics is the use of statistical modelling and / or machine learning methods to measure the future. In short, it is one of the data mining techniques for predictive analysis that focuses on creating a predictive model for the future by extracting relationships from the data.