I am working at Anadolu Insurance Company as a data analyst. My main responsibilities are collaborating with the IT teamto develop and optimize data warehouse and participating in IT projects as a data analyst. I am really interested in data science and my goal is to become a data sicentist.
The lecturer, Mine Çetinkaya, introduces datasciencebox.org which is a website created by herself to teach data science. She mentions her ideas about the three main aspects of teaching which are content, pedagogy and infrastructure. She also explains teaching tools and methods used in the website. Although there are many resources and courses related to data science, I think that most of these resources do not contribute sufficiently to the students. In this context I think her education approach is quite different and eye opener.
In this video, the lecturer gives information about new machine learning techniques on R in order to make better pricing for insurance industry. Orthodox methods for insurance pricing are Generalized Linear Models (GLMs) and Generalized Additive Models (GAMs). The new machine learning techniques shows that the old methods are not so bad but more pricise predictions could be made by Gradient Boosting Machine (GBM) method which is a machine learning technique for regression and classification problems.
The lecture is about sports analytics. The training data, collected from the training session of a team, is used in order to planning upcoming drills. The data shows details of exercise such as the distance of players run, their acceleration and so on. The lecturer mentions that random effects model and multi level model are used for predicton because the linear regression model is not appropriate for team data. The output of the model is estimates about some metrics such as the distance, the average heart rate, the average metabolic power and so on. This video is not directly related to insurance industry but if this model is extended to include an injury model it would be a good resource for personal accident insurance predictions.
A Shiny App Used to Predict Training Load in Professional Sports
The main focus of the study is immune cell frequencies that are a key indicator of human health and disease. By using standardized flow cytometry and questionnaire the dataset, which includes biomolecular outputs and also demographical and socio-economical charecteristics such as income education diet depression and so on, is genereated. The investigation is the relationship between demographical dataset and immune cell frequencies. The lecturer mentions that they have developed a package in R, MMI, and used different regression models for this investigation. The outputs of this study can be used as a component in health insurance pricing.