Predictive Analytics

Predictive Analytics

Regardless of kinds of industry, demand and risk predictions of operation are significant issues to the operators. Because predictions can make the operators to prepare and handle the market and the problems in advance. In addition, it has become possible to predict accurately by using data for prediction. Thus, it is growing that the interest and importance of predictive analytics based on data. In our laboratory, we are contributing to advance of predictive analytics by studying these topics; 1. Development of predictive models using field data and related other data and 2. Establishment of services and platforms for effective operation by prediction methodologies and result.

  • Development of predictive models using field data and related other data
    • Studying prediction methodology based on machine learning for facility predictive maintenance by state monitoring and prediction using facility data
    • Drawing the optimal resources level from the simulation based on the prediction result of analytic and predictive algorithm that developed in the past; and, prediction of various situation that can occur in the field by inferring the operating prediction distribution and making the predictive process model from data.
    • Study the predictive methodology according to the wider inference of field situations under the conditions that was done before.
    • Development of multi-layered LSTM predictive analytics methodology for exact medium and long-term prediction of resource pattern and demand by using data.

  • Establishment of services and platforms for effective operation by predictive methodologies and results
    • Development and construction the services of facility state prediction based on monitoring and facility data pattern analytics; the services use the data that was collected in real time from IoT sensors.
    • Optimization of operating plans using reinforcement learning based on using operating the facility data comprehensively including the predictive results of the uncertain outdoor operating facility states.
    • Developing the data analytic and predictive platform that can be used by domain expert easily with introduction of the methodologies like process mining, statistical methods, simulation, and artificial intelligence for operating big data analytics.

Process Analytics

Process Analytics

Process analytics is an analytic methodology that is used for understanding, managing, and improving the business processes by defining and explaining each factor and relationship between the factors. When analyzing the past data by the method, it can get some information that can know only from the field; the information cannot recognize from the related documents. According to analytics based on the field’s situations, it can solve the problems practically. Thus, the laboratory studies about 1. Process visualization, diversification of the analytic viewpoint, and methods of drawing processes using data, 2. Conformance analytics of the process model and method of processing the missing an anomaly data using the results, 3. Process optimization and simulation according to the drawn process and the analytic results. According to the topics, the lab discovers and deals with the processes, and researches the application plans used the drawing results.

  • Process visualization, diversification of the analytic viewpoint, and methods of drawing processes using data
    • Defining the events that exist in the process, deciding the fixed factors used for process analytics by standardizing the components of the meta data, and drawing the optimal fixed process
    • Methodology developments that analyze and draw the process model immediately when urgent changes occur in the field that the real process are run.
    • Studying the algorithm of analyzing and visualizing the process that helps understand the intuitive expression of the analytic results easily by the domain experts and that connect with the fields in real time.

  • Conformance analytics of the process model and method of processing the missing and anomaly data using the results
    • Development of an algorithm that gives stable comparison by simplifying the data based on the statistics with maintaining the equivalence of the methodologies of process conformance analytics used before.
    • Processing the missing and anomaly data according to the order and connection of the operating data by using process models.
    • Making a processing methodology that improve the accuracy of the processing the missing or anomaly data with process analytics, predictions, and log data generated results of the outdoor environment connected of field data.

  • Process optimization and simulation according to the drawn process and the analytic results
    • Flexible process operation support to decide the needs of new scheduling and schedule again by reflecting the log data in real time.
    • Assistant of drawing the optimal operating environment for optimization of process by development of a simulation tool that assume and apply the decision making list based on process mining.
    • Managing plans automatically and improving flexibility by development of scheduling methodology based on log data with artificial intelligence or reinforcement learning

Data Infrastructure

Data Infrastructure

Recently, according to growing the data volume, the number of needed resources is increasing for processing it effectively. At the same time, the processing methodologies is complexifying, and thus other domain experts have become difficult to deal with the method without the knowledge of data processing. However, study and application of big data are hot issues in the society, so the enterprises are increased that are interested in and provide platforms that help use the huge data not bounded by time. In addition, clouds are used for providing what they need, and the users can make an attempt using of data easily with low price. Through the needs of data infrastructure for using data is increasing in the industry, this laboratory studies 1. Building a bigdata platform for providing optimal environments of using data, and 2. Development of cloud service that used in data processing and analytics.

  • Building a big data platform for providing optimal environments of using data
    • Studying the event log distributed replay algorithm that can play the larger log data that couldn’t replay in the web browsers before.
    • At data lake that is extended than data warehouse, developing automatic services that can check, extract, search, process, and analyze freely without users’ efforts.

  • Development of cloud service that used in data processing or analytics
    • Establishment of the service that provide a simulation analytic tool based on the operating big data through the web.
    • Predictive analytics service development on cloud that makes easy to use for anyone who wants data predictive analytics by making intuitive evaluation index of data analytics and standardizing the processes of the predictive analytics.
    • Building the base of providing a customizing analytic service that needs short time and is easy by designing the standardized data analytic service based on machine learning and modularizing the model structures for customizing.