Advances in Data Science and Adaptive Analysis

Recommender Systems (RSs) have become an essential part of most e-commerce sites nowadays. Though there are several studies conducted on RSs, a hybrid recommender system for the real state search engine to find appropriate rental apartment taking users preferences into account is still due. To address this problem, a hybrid recommender system is proposed in this paper constructed by two of the most popular recommendation approaches — Collaborative Filtering (CF), Content-Based Recommender (CBR). CF-based methods use the ratings given to items by users as the sole source of information for learning to make a recommendation. However, these ratings are often very sparse in applications like a search engine, causing CF-based methods to degrade accuracy and performance. To reduce this sparsity problem in the CF method, the Cosine Similarity Score (CSS) between the user and predicted apartment, based on their Feature Vectors (FV) from the CBR module is utilized. Improved and optimized Singular Value Decomposition (SVD) with Bias-Matrix Factorization (MF) of the CF model and CSS with FV of CBR constructs this hybrid recommender. The proposed recommender was evaluated using the Statistical Cross-Validation consisting of Leave-One-Out Validation (LOOCV). Experimental results show that it significantly outperformed a benchmark random recommender in terms of precision and recall. In addition, a graphical analysis of the relationships between the accuracy and error minimization is presented to provide further evidence for the potentiality of this hybrid recommender system in this area.

This paper presents the automated tool for the classification of text with respect to predefined categories. It has always been considered as a vital method to manage and process a huge number of documents in digital forms which are widespread and continuously increasing. Most of the research work in text classification has been done in Urdu, English and other languages. But limited research work has been carried out on roman data. Technically, the process of the text classification follows two steps: the first step consists of choosing the main features from all the available features of the text documents with the usage of feature extraction techniques. The second step applies classification algorithms on those chosen features. The data set is collected through scraping tools from the most popular news websites Awaji Awaze and Daily Jhoongar. Furthermore, the data set splits in training and testing 70%, 30%, respectively. In this paper, the deep learning models, such as RNN, LSTM, and CNN, are used for classification of roman Urdu headline news. The testing accuracy of RNN (81%), LSTM (82%), and CNN (79%), and the experimental results demonstrate that the performance of the LSTM method is state-of-art method compared to CNN and RNN.

The real-time data analysis requires an integrated approach to know the last known state of variables of a concept under monitoring. Thereby, the Internet-of-Thing (IoT) devices have provided alternatives to address distributed data collection strategies. However, the autonomy of IoT devices represents one of the main challenges to implement the collecting strategy. Battery autonomy is affected directly by the energy consumption derived from data transmissions. The Data Stream Processing Strategy (DSPS) is an architecture oriented to the implementation of measurement projects based on a measurement and evaluation framework. Its online processing is guided by the measurement metadata informed from IoT devices associated with a component named Measurement Adapter (MA). This paper presents a new data buffer organization based on measurement metadata articulated with online data filtering to optimize the data transmissions from MA. As contributions, a weighted data change detection approach is incorporated, while a new local buffer based on logical windows is proposed for MA. Also, an articulation among the data buffer, a temporal barrier, and data change detectors is introduced. The proposal was implemented and released on the pabmmCommons library. A discrete simulation on the library is here described to provide initial applicability patterns. The data buffer consumed 568 Kb for monitoring 100 simultaneous metrics. The online estimation of the mean and variance based on the Statistical Process Control consumed 238 ns. However, as a limitation, other scenarios need to be addressed before generalizing results. As future work, new alternatives to filter noise online will be addressed.

During the past few years, deep learning (DL) architectures are being employed in many potential areas such as object detection, face recognition, natural language processing, medical image analysis and other related applications. In these applications, DL has achieved remarkable results matching the performance of human experts. This paper presents a novel convolutional neural networks (CNN)-based approach for the detection of breast cancer in invasive ductal carcinoma tissue regions using whole slide images (WSI). It has been observed that breast cancer has been a leading cause of death among women. It also remains a striving task for pathologist to find the malignancy regions from WSI. In this research, we have implemented different CNN models which include VGG16, VGG19, Xception, Inception V3, MobileNetV2, ResNet50, and DenseNet. The experiments were performed on standard WSI slides data-set which include 163 patients of IDC. For performance evaluation, same data-set was divided into 113 and 49 images for training and testing, respectively. The testing was carried out separately over each model and the obtained results showed that our proposed CNN model achieved 83% accuracy which is better than the other models.

Multiple regression is not reliable to recover predictor slopes within homogeneous subgroups from heterogeneous samples. In contrast to Monte Carlo analysis, which assigns completely to the first-specified predictor the variation it shares with the remaining predictors, multiple regression does not assign this shared variation to any predictor, and it is sequestered in the residual term. This unassigned and confounding variation may correlate with specified predictors, lead to heteroscedasticity, and distort multicollinearity. I develop and test an iterative, sequential algorithm to estimate a two-part series of weighted least-square (WLS) multiple regressions for recovering the Monte Carlo predictor slopes in three homogeneous subgroups (each generated with 500 observations) of a heterogeneous sample $(n=1,500)$. Each variable has a different nonnormal distribution. The algorithm mines each subgroup and then adjusts bias within it from 1) heteroscedasticity related to one, some, or all specified predictors and 2) “nonessential” multicollinearity. It recovers all three specified predictor slopes across the three subgroups in two scenarios, with one influenced also by two unspecified predictors. The algorithm extends adaptive analysis to discover and appraise patterns in field research and machine learning when predictors are inter-correlated, and even unspecified, in order to reveal unbiased outcome clusters in heterogeneous and homogeneous samples with nonnormal outcome and predictors.