**Author details**

Ali Soofastaei Artificial Intelligence Center, Vale, Brisbane, Australia

\*Address all correspondence to: ali@soofastaei.net

© 2019 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/ by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

*Advanced Analytics and Artificial Intelligence Applications*

AI solutions are strongly related to background processes and collected datasets. In other words, the reliability of AI applications depends on the quality of historical information. AA applications usually deal with critical challenges related to security, maintenance, operations, energy consumption, and safety. Dissatisfaction with prediction, optimization, or decision-making algorithms may lead to negative outcomes and discourage users from relying on AA approaches such as AI systems.

AI can help improve the accuracy and reliability of prediction and optimization of industrial applications. However, interpreting the results is a significant challenge for experts and managers when using AI to solve business problems. A practical solution for industry may be to train experts, specialists, and managers to operate the analytics and provide root cause analysis for anomalies. This implies that during the development of applications, data scientists should work with experts and managers to include domain knowledge in algorithm expansion processes and ensure that models can adaptively learn and accumulate knowledge.

Predictive modeling is the term used for the process of utilizing data mining and probability to forecast future outcomes. A predictive (forecasting) model uses several independent variables or predictors that are likely to influence the desired dependent variable (forecasting output). Once data have been collected for the relevant predictors, a statistical algorithm is deployed. This algorithm may be a simple linear equation or a sophisticated ML algorithm such as a neural network. Predictive modeling mainly overlaps with the field of ML, and many of the algorithms utilized

Predictive modeling is often associated with weather forecasting, online advertising, and marketing. However, it also has applications in mining engineering. One of the most frequently overlooked challenges of the forecasting model is obtaining suitable data to apply when creating algorithms. Data collection and preparation is the most challenging step in developing a predictive model, and it is essential to locate the best predictors to feed into the model. A descriptive analysis of the data and data treatment, including missing values and outlier fixing, is a crucial task that consumes most of the time needed in predictive modeling. Once the data have been collected, the next step is to select an appropriate model. Linear regressions are among the most accessible models for predictive algorithms, but other multifaceted AI models are available. The complexity of the model does not guarantee the performance of the prediction. Model selection should be considered in relation to data availability and quality and the forecasting period. After modeling, a production estimation should be provided to measure the

Some well-known predictive modeling methods widely used in industrial appli-

Many different practical optimization methods have been used in critical industries. Generally, the aim of optimization is to increase productivity, energy and cost

cations are regression, time series algorithms, deep learning, and ML.

in forecasting models are found in the context of ML and AI.

**2.3 High reliability**

**2.4 Interpretability**

**3. Predictive models**

accuracy of the model.

**4. Optimization methods**

**4**
