DeutscheAktuarvereinigung's repositories
claim_frequency
GLM, Neural Network and Gradient Boosting for Insurance Pricing, Part 1: Claim Frequency
Mortality_Modeling
Multi-Population Mortality Modeling With Neural Networks
insurance_scr_data
How to Work With Comprehensive Internal Model Data for Three Portfolios
Deriving-NHANES-data-set-CDC-for-mortality-analysis
Deriving of a NHANES-data set (CDC) for a mortality analysis
Data_Science_Challenge_2020_Betrugserkennung
In this notebook we take a look at a relevant project that is frequently encountered by insurers: Fraud Detection. For this purpose we use a car data set from a public source and will show the necessary steps to establish an automated fraud detection.
Data-Science-Challenge2021_Explainable-Machine-Learning
The notebook on the main topic of interpretable machine learning is a descriptive and instructive analysis of a car data set from a public source.
Data_Science_Challenge_2020_Berufsunfaehigkeit
The study Machine-Learning Methods for Insurance Applications is dedicated to the question of how new developments in the collection of data and their evaluation in the context of Data Science in the actuarial world can be utilized. The results of the study are based on the R language, so the first goal of this work is to reproduce the calculations described in the Jupyter notebook in the Python programming language and to compare the results with those of the study authors. Besides these presented methods we continue to work on a random forest. Therefore, our second goal is the development of an artificial neural network, which has at least a similar quality compared to the other machine learning methods.
ADS_Use_Cases
Notebooks etc. for Actuarial Data Science use cases
Data_Science_Challenge_2022_Python-Notebook_zur_Erstellung_von_Schadenhaeufigkeitsmodellen
In this Python notebook, based on a large French. The results are compared and the interpretability of the models is analyzed and evaluated with SHAP and PDP plots. In addition, the four tools TPOT, Auto-Sklearn, H2O and FLAML are tested or used.
Impact_of_the_COVID-19_Pandemic
Modeling and Forecasting using Affectedness Variables