Large scale tensor regression using kernels and variational inference

We outline an inherent flaw of tensor factorization models when latent factors are expressed as a function of side information and propose a novel method to mitigate this. We coin our methodology kernel fried tensor (KFT) and present it as a large-scale prediction and forecasting tool for high dimensional data. Our results show superior performance against LightGBM and Field aware factorization machines (FFM), two algorithms with proven track records, widely used in large-scale prediction. We also develop a variational inference framework for KFT which enables associating the predictions and forecasts with calibrated uncertainty estimates on several datasets.

Forecasting of Air Maximum Temperature on Monthly Basis Using Singular Spectrum Analysis and Linear Autoregressive Model

In this research, the singular spectrum analysis technique is combined with a linear autoregressive model for the purpose of prediction and forecasting of monthly maximum air temperature. The temperature time series is decomposed into three components and the trend component is subjected for modelling. The performance of modelling for both prediction and forecasting is evaluated via various model fitness function. The results show that the current method presents an excellent performance in expecting the maximum air temperature in future based on previous recordings.

Research on Abnormal Detection Method of Data Flow Trend in Oil Drilling Engineering Early Warning

In this paper, by collecting and analysing the domestic and foreign oil drilling accident data and early warning technology, combined with the oil drilling process prediction and forecasting process deficiencies. Based on theoretical analysis and analysis of five typical accident modes in actual drilling rig engineering, the early warning method and train of thought of oil drilling engineering are established, and the index of early warning of accident is given. The sensitivity of various accident prediction indexes to the corresponding accidents is studied. On the basis of analysing the forecast signal and processing, the comprehensive model of accident early warning under multi-objective condition is established.

Prediction and Forecasting of Maximum Weather Temperature Using a Linear Autoregressive Model

This paper investigates the autoregressive (AR) model performance in prediction and forecasting the monthly maximum temperature. The temperature recordings are collected over 12 years (i.e., 144 monthly readings). All the data are stationaries, which is converted to be stationary, via obtaining the normal logarithm values. The recordings are then divided into 70% training and 30% testing sample. The training sample is used for determining the structure of the AR model while the testing sample is used for validating the obtained model in forecasting performance. A wide range of model order is selected and the most suitable order is selected in terms of the highest modelling accuracy. The study shows that the monthly maximum temperature can accurately be predicted and forecasted using the AR model.