Having forecast of real estate sales done correctly is very important for balancing supply and demand in the housing market. However, it is very difficult for housing companies or real estate professionals to determine how many houses they will sell next year. Although this does not mean that a prediction plan cannot be created, the studies conducted both in Turkey and different countries about the housing sector are focused more on estimating housing prices. Especially the developing technological advances allow making estimations in many areas. That is why the purpose of this study is both to provide guiding information to the companies in the sector and to contribute to the literature. In this study, a 124-month data set belonging to the 2008 (1) - 2018 (4) period has been taken into account for total housing sales in Turkey. In order to estimate the time series of sales, ARIMA (Auto Regressive Integrated Moving Average as linear model), LSTM (Long Short-Term Memory as nonlinear model) has been used. As to increase the estimation, a HYBRID (LSTM and ARIMA) model created has been used in the application. When MAPE (Mean Absolute Percentage Error) and MSE (Mean Squared Error) values ​​obtained from each of these methods were compared, the best performance with the lowest error rate proved to be the HYBRID model, and the fact that all the application models have very close results shows the success of predictability. This is an indication that our study will contribute significantly to the literature.

Earning via real-time predictions with the experience in the visible trend directions of an investment instrument in the past requires a different perspective on charts. Indicators and formations within the scope of technical analysis constitute the most significant basis of this perspective. Those who can generate a high income in financial markets and even be more successful than large companies are actually the ones interpreting the data in a different way. In this study, a model which had never been encountered in the literature before, was designed through a different perspective on the same data, enabling the movements of an investment element over the 2D candlestick chart to be recognized as a "Buy-Sell" object respectively and to decide on the trend direction as a result. The model is trained by state-of-the-art, real-time object detection system (You Only Look Once) YOLO; for the training, one-year candlestick charts belonging to the stocks traded on Borsa İstanbul (BIST) between 2000-2018 were used. The model, which can make a "Buy-Sell" decision without the need for an additional time series except for the views on the visual candlestick charts, is promising in terms of its successful predictions. Its ultimate aim is to provide a foresight strengthening the "Buy-Sell" decisions to be made in the decision-making process following the other basic and technical analyses in addition to its stand-alone use in making investment decisions. The effect of this foresight on the success can clearly be seen on the test results received. In the results, the model was found to be successful by 85% while a 100% profit was generated. Besides, the model can be used for all the time series for which candlestick charts can be created.

The fundamental purpose of every investor making investments in financial fields, is to make profit by buying an investment instrument at a low price and selling the same at a higher price. In this study, within the framework of the aforementioned standpoint, an effective “Trading” decision support model was designed, which can be used for stock market analyses, parity analyses, index analyses, and for the stock analyses of other stock exchanges, briefly for all investment instruments for which candlestick charts are created. An innovative model design was achieved through a bilateral perspective with both financial and scientific aspects by designing these models that operated on a pattern detection basis. The study incorporated the use of 2D candlestick charts of the BIST stocks. The charts were labeled in two separate data sets. Initially, 10,000 pieces of data were labeled on 550 2D candlestick charts, which were trained with YoloV3 Data Group-1 (DG-1) [1]. Subsequently, the data set was increased to 20,000 pieces. Out of this set of 20,000 labeled data prepared, 10,000 labeled data were picked randomly. The newly-created set of 10,000 labeled data was named DG-2, which was trained with the YoloV3, YoloV4, Faster R-CNN, SDD algorithms. An assessment was made regarding the performance results obtained following the trainings implemented for these four chosen algorithms. For the aforementioned assessment, three different scenarios were created, and out of all these scenarios, the YoloV3 DG-2 algorithm, which was trained with an improved data set, was observed to be most successful one. As a result of the comparative test scenarios, the YoloV3 DG-2 model achieved a pattern recognition success of 98%. On the other hand, it was also observed to have achieved a prediction success of 100%, while bringing in a return by 89.94%, regarding the object class detected. In addition, no additional parameters were used in this observed gain success. Consequently, the YoloV3 DG-2, determined as the final model, could be implemented as a decision support model for all investment instruments for which a candlestick chart can be created.

The increased speed rates and ease of access to the Internet increase the availability of devices with Internet connections. Internet users can access many devices that they are authorized or not authorized. These systems, which detect whether users have unauthorized access or not, are called Intrusion Detection Systems. With intrusion detection systems, users' access is classified and it is determined whether it is a normal login or an anomaly. Machine learning methods undertake this classification task. In particular, Boosting algorithms stand out with their high classification performance. It has been observed that the Gradient Boosting algorithm provides remarkable classification performance when compared to other methods proposed for the Intrusion Detection Systems problem. Using the Python programming language, estimation was made with the Gradient Boost, Adaboost algorithms, Catboost, and Decision Tree and then the model was explained with SHAPASH. The goal of SHAPASH is to enable universal interpretation and comprehension of machine learning models. Providing an interpretable and explainable approach to Intrusion Detection Systems contributes to taking important precautions in the field of cyber security. In this study, classification was made using Boosting algorithms, and the estimation model created with SHAPASH, which is one of the Explainable Artificial Intelligence approaches, is explained.

In data mining, the process of grouping similar data among each other in an evaluation made by the same kind of data is called clustering. Clusters are usually computed by algorithms that take advantage of relational or remote neighborhood relations to each other. A large number of clustering methods are used in practice. In this study, the differences between open source software used in data mining and results obtained by clustering methods of Oracle, Knime, Weka and RapidMiner programs are tabulated. As a source for this study, clustering analysis based article presentations by M. Turanlı, Ü.H.Özden and S.Türedi have been handled. For the evaluation, in 2006, economic similarities of 29 countries from candidate and member countries from the EU in terms of proprtion of inflation, Gross National Product, Internet Usage, Unemployment, Lifetime Education Indicators and Imports and Exports were tried to be clustered. Clustering operations were performed separately as 2-3-4 clusters in each program. In the clustering process, the euclidean distance is set for calculation and the number of iterations is 99 in each program. The obtained results were written in tabular form and the results were interpreted.

Öz Bir verinin bir dizgi içerisinde veya bir gen yapısının bir DNA gen dizilimi içerisinde arama işleminin gerçekleştirilmesi için çeşitli algoritmalar kullanılmaktadır. Kullanılan bu algoritmalardan bazıları bize mutlak eşleşme olmadığı durumlarda olumsuz dönüt vermekte, bazıları ise “bunu mu arıyorsunuz” diye alternatifler sunmaktadır. Her iki algoritma da genel amaçlı PC’lerde saniyeler süren işlemler sonucunda bize dönüt verebilmektedir. Bu çalışmada bize hem mutlak eşleşmeyi hem de hedef dizgi içinde yüzdelik eşleşme oranlarının gerçekleştiği konumu veren FPGA çiplerine yönelik yüksek performanslı bir donanım modülü tasarlanmıştır. Geliştirilen modülün veri işleme hızı farklı PC’lerle karşılaştırılmış ve 2300 kata kadar daha hızlı arama gerçekleştirdiği karşılaştırma sonuçlarından elde edilen veriler ile doğrulanmıştır.