Development of an experiment-split method for benchmarking the generalization of a PTM site predictor: Lysine methylome as an example

Many computational classifiers have been developed to predict different types of post-translational modification sites. Their performances are measured using cross-validation or independent test, in which experimental data from different sources are mixed and randomly split into training and test sets. However, the self-reported performances of most classifiers based on this measure are generally higher than their performances in the application of new experimental data. It suggests that the cross-validation method overestimates the generalization ability of a classifier. Here, we proposed a generalization estimate method, dubbed experiment-split test, where the experimental sources for the training set are different from those for the test set that simulate the data derived from a new experiment. We took the prediction of lysine methylome (Kme) as an example and developed a deep learning-based Kme site predictor (called DeepKme) with outstanding performance. We assessed the experiment-split test by comparing it with the cross-validation method. We found that the performance measured using the experiment-split test is lower than that measured in terms of cross-validation. As the test data of the experiment-split method were derived from an independent experimental source, this method could reflect the generalization of the predictor. Therefore, we believe that the experiment-split method can be applied to benchmark the practical performance of a given PTM model. DeepKme is free accessible via https://github.com/guoyangzou/DeepKme.

Eye Tracking and an A/B Split Test for Social Media Marketing Optimisation: The Connection between the User Profile and Ad Creative Components

In the era of digitally dominated competition, where the effectiveness of Facebook ads prompts the social media marketing strategy, new opportunities arise for most industries. In this context, this study aims to discover and summarize the key optimisations regarding the creative ad components needed to run highly personalized ads based on a user profile. The participants were Facebook users between the ages of 18 and 65+ who were interested in dental services. Qualitative methods were adopted in order to discover suitable options for professionals to grow their business. In the first stage, the A/B split test using the Facebook Ads manager labelled the most effective creative component. In the second stage, an eye-tracking experiment generated 30 heatmaps that showed the differences between the segments. The results show solutions for attracting users by increasing the level of personalization of the ads. They are more beneficial for social media campaigns aimed at brand awareness targeting women and showing them a happy human face. When the target audience is men, technical details are preferred in order for the ad to become more attractive for them. This study enriches the literature and empowers professionals to deploy social media marketing growth strategies to attract users and make them convert to their full potential.

Fracture Behaviour of Concrete with Reactive Magnesium Oxide as Alternative Binder

This research evaluates the fracture behavior of concrete with reactive magnesium oxide (MgO). Replacing cement with MgO is an attractive option for the concrete industry, mainly due to sustainability benefits and reduction of shrinkage. Four different MgO’s from Australia, Canada, and Spain were used in the concrete mixes, as a partial substitute of cement, at 5%, 10%, and 20% (by weight). The fracture toughness (KI) intensity factor and the stress–strain softening parameters of the wedge split test were evaluated after 28 days. The experimental results showed that the replacement of cement with MgO reduced the fracture energy between 13% and 53%. Moreover, the fracture energy was found to be correlated with both compressive strength and modulus of elasticity. A well-defined relationship between these properties is important for an adequate prediction of the non-linear behavior of reinforced concrete structures made with partial replacement of cement with MgO.

Fracture bi-linear model parameters and brittleness of high-performance concretes for paving: why fatigue tests matters and cannot be spared?

ABSTRACT: Fracture tests on two high-performance concrete for pavements were performed using the wedge-split test. The results allowed perceiving the same specific fracture energy for both concrete. However, the crack opening during the tests as well as the characteristic length of the materials resulted in roundly distinct behaviors in terms of brittleness. Previous fatigue studies for both concretes are then fairly readdressed pointing out the unavoidable needs for studying the concrete brittleness as a parametric way for selecting suitable concrete proportions and materials as well as to reach its microstructural behavior in order to acquire an appropriate judgment about its fatigue performance.

PREDIKSI INDEKS PRESTASI MAHASISWA YANG BERKULIAH SAMBIL BEKERJA DI UNIVERSITAS ADVENT INDONESIA DENGAN MENGGUNAKAN METODE DECISION TREE C4.5 DAN SMOTE

Pendidikan tinggi adalah salah satu cara agar mendapat pekerjaan lebih mudah, hal tersebut disebabkan karena melalui pendidikan individu tersebut mampu meningkatkan kuliatas sumber daya manusia pada zaman ini. Namun biaya pendidikan yang tinggi sangat mahal sehingga individu yang ingin berkuliah harus juga bekerja disaat yang bersamaan, maka penelitian ini bertujuan untuk memprediksi indeks prestasi mahasiswa yang berkuliah sambil bekerja di Universitas Advent Indonesia. Dari hasil penelitian ini terdapat 8 atribut yang berpengaruh dalam memprediksi indek prestasi mahasiswa di Universitas Advent Indonesia yaitu Departemen Pekerjaan, Jam Kerja, Jurusan, Jenis Kelamin, Tempat Tinggal, Usia, Jumlah SKS dan Indeks Prestasi. Metode yang digunakan dalam penelitian ini adalah Decision Tree C4.5 yang diimplementasikan pada program WEKA dengan algoritma J48. Penelitian ini juga menggunakan algoritma SMOTE (Synthetic Minority Oversampling Technique) untuk menyeimbangkan jumlah data pada kelas minor. Root teratas dari penelitian ini adalah Jenis Kelamin yang mempengaruhi indeks prestasi mahasiswa di Universitas Advent Indonesia. Algoritma SMOTE pada penelitian ini berguna untuk membantu menaikan hasil dari penelitian ini sebesar 7-8% bisa dilihat dari hasil akurasi pengujian cross validation 10 folds adalah 63.6672%, kemudian rata-rata hasil dari precision dan recall adalah 0.621 dan 0.637. Sementara untuk hasil akurasi dari split test 70:30 adalah 62.7955%, kemudian rata-rata hasil dari precision dan recall adalah 0.621 dan 0.628. Jika dibandingkan dengan penggunaan algoritma decision tree C4.5 saja maka, akurasi dari pengujian cross validation 10 folds adalah 55.5044%, dengan rata-rata hasil dari precision dan recall adalah 0.545 dan 0.555. Sementara hasil akurasi dari split test 70:30 adalah 55.2995% dengan rata-rata hasil dari precision dan recall adalah 0.554 dan 0.553. Hasil analisa menggunakan confusion matrix serta kurva ROC dengan hasil dari 0.688­ menjadi 0.756, yang berada dalam rentang 0.70 – 0.80 yang masuk kedalam tingkat diagnose fair classification. Dapat disimpulkan bawa terdapat pengaruh berkuliah sambil bekerja yang cukup kuat terhadap indeks prestasi mahasiswa. Dengan urutan atribut dari yang paling atas adalah Jenis Kelamin, Jumlah SKS, Jurusan, Umur, Departemen Kerja, Jam Kerja dan Tempat Tinggal.

Numerical Brazilian split test of pre-cracked granite with randomly distributed micro-components

Purpose The purpose of this paper is to investigate the mechanical behavior and propagation of cracks of numerical granite samples through the Brazilian split test and to provide a reference for predicting the behavior of real granite samples. Design/methodology/approach The numerical models of granite containing two fissures are established using the parallel bond model (PBM) and the smooth joint model (SJM) in PFC2D. The peak stresses, number of cracks and anisotropic ratios are obtained to study the influence of the mineral composition and the angle of inclination of rock bridge on the strength, failure mode and deformation characteristics. Findings The numerical results obtained show that the mineral composition has a marginal influence on the peak stress. When the angle of inclination of rock bridge β increases, the peak stress drops to its minimum value at β = 90° and then gradually increases to a relatively low level. The behavior of cracks falls into three categories based on the distribution of cracks. By analyzing the stress–strain curve and the process of crack propagation for sample No. 4 with β = 60°, it is found that the process of failure can be divided into four stages and tensile cracks dominate. The anisotropic ratios of peak stress and a number of cracks obtained show that the peak stress is low anisotropic and the number of cracks is medium anisotropic. Originality/value This paper presents a numerical simulation method to analyze mechanical behavior and propagation of cracks under different conditions. The proposed method and the results obtained are useful for predicting the behavior of real granite samples in laboratory and engineering projects.

Sentiment analysis of customer reviews in zomato bangalore restaurants using random forest classifier

Introduction: Natural Language Processing is one part of Artificial Intelligence and Machine Learning to make an understanding of the interactions between computers and human (natural) languages. Sentiment analysis is one part of Natural Language Processing, that often used to analyze words based on the patterns of people in writing to find positive, negative, or neutral sentiments. Sentiment analysis is useful for knowing how users like something or not. Zomato is an application for rating restaurants. The rating has a review of the restaurant which can be used for sentiment analysis. Based on this, writers want to discuss the sentiment of the review to be predicted. Method: The method used for preprocessing the review is to make all words lowercase, tokenization, remove numbers and punctuation, stop words, and lemmatization. Then after that, we create word to vector with the term frequency-inverse document frequency (TF-IDF). The data that we process are 150,000 reviews. After that make positive with reviews that have a rating of 3 and above, negative with reviews that have a rating of 3 and below, and neutral who have a rating of 3. The author uses Split Test, 80% Data Training and 20% Data Testing. The metrics used to determine random forest classifiers are precision, recall, and accuracy. The accuracy of this research is 92%. Result: The precision of positive, negative, and neutral sentiment is 92%, 93%, 96%. The recall of positive, negative, and neutral sentiment are 99%, 89%, 73%. Average precision and recall are 93% and 87%. The 10 words that affect the results are: “bad”, “good”, “average”, “best”, “place”, “love”, “order”, “food”, “try”, and “nice”.