Predictive analytics architecture for iot focused big data

Technological developments increase possibilities and capabilities of researchers day by day. Data-centered business processes gain importance by the developments in hardware, software and computational techniques in artificial intelligence and related fields. This situation leads to requirement of managing data storage, data transfer and data processing. Spread of internet of things approach, where each device in the production areas is seen as a data source, provides solutions for the creation of large amounts of data and the production of real-time or near-real-time outputs. Many studies point out various problems and solutions under the big data and internet of things by using cloud technologies. These developments bring paradigm shifts and a tendency in institutions to position key people who can use big data instruments and operate analytical processes in industrial environments. Importance of programming skills for users decrease, while importance of ability to use software tools with no code or low code increase.There are many integrated processes from obtaining data to reporting by business intelligence or operational dashboarding tools. Goals such as increasing digitalization, traceability and analytical capability are among priorities of the organizations. In the context of analytical capabilities, executing machine learning algorithms, training models, deploying the trained models as software services and reporting their outputs to various digital environments also have to work in integration with big data processes. This thesis, presents a predictive analytics architecture for big data with the focus of the internet of things, taking into account cloud-native environment requirements. Open source softwares and software development platforms are also indicated, along with functional breakdowns over state of the art to reveal the relevant architectural structure. In addition to concepts such as data lake and lakehouse, design approaches to streaming and stream processing are explained theoretical and exemplified by software tools. Similarly, concepts and software tools related with Internet of Things approach are also exemplified. The presented architecture includes components that cover the entire story of the data and can be benefitted from by software developers, key users and end users.