Some examples of programming and data science projects I worked on.
To better understand the weather in my hometown, I applied classical time series methods (ARIMA) and deep learning (LSTM) to forecast the monthly cumulated rainfall and the average temperature in Trier. Here are the results of my most accurate prediction using LSTM:
with the following accuracy in the predictions:
MAPE | RMSE | |
---|---|---|
Rainfall | 3.48% | 2.48 |
Temperature | 2.02% | 1.27 |
I wanted to apply the methods from my weather forecast project to another classic time series. So I used ARIMA and LSTM to forecast stock prices in the example of Microsoft's closing stock prices. Here are the results of my most accurate prediction using LSTM:
with the following accuracy in the predictions:
MAPE | RMSE |
---|---|
0.72% | 1.39 |
I adjusted a scraper for glassdoor.com to the german glassdoor page and gathered data of around 1000 data science job offers to support my job search. For example, I wanted to find the most requested keywords in the job descriptions and created the following graph:
A project of an internship at a major insurance company in 2013 that made savings in reinsurance premium possible by controlling the risk of a change of their reinsurance retention. Here are some results for realistic and synthetically generated data:
An overview of machine learning, artificial intelligence and big data methods and its application will be created in this repo.Overview of the applications and used machine learning methods in official statistics from the meta analysis of Beck, Dumpert, Feuerhake, (2018):