Context People living in Switzerland have to pay more for healthcare than almost anybody else worldwide. The USA are an exception (https://data.oecd.org/healthres/health-spending.htm). Obviously this is a very sensitive and intensely debated issue. There are 54 active insurance companies in the market, they provide the mandatory insurance with different tariffs for different models and geographical regions. In general the premiums are rising, but sometimes there are big changes for single tariffs from year to year which allows consumers to save significantly. It is not very transparent how these single premiums are calculated. They are communicated yearly by the end of September and only then all premiums for the next year will be known. But if these figures could be predicted earlier, the more time consumers would have to plan for a possible change. Content The statistics about the Swiss healthcare system are published by the government. They consist of information about the insurers, regions, healthcare providers and demographic information. It takes some time to get the complete data, actually published are the figures for 2016 (including the premiums for 2017). The information is provided in Excel-files, normally in pivoted form. The Excels were loaded from several webpages of the Swiss Federal Office of Public Health. In a first step the worksheets (around 140) were unpivoted and their dimension information was standardized: After this data with matching dimensionality was consolidated and written into a database. The tables containing relevant dimensions were joined and pivoted into one large table that builds the basis for our analysis. [Swiss Federal Office of Public Health][1] Inspiration Apart from finding a precisely working ML model we would love to learn how the premiums are calculated. More precisely: which factors have the biggest influence on the final price a customer has to pay. So it’s not only about machine learning, but feature engineering plays a big role in our challenge. Depending on the resulting factors it will be possible to get the corresponding actual data and make a prediction about the expected premiums for 2019. Our table contains some category information (which can be ignored) and 200 describing numeric columns. Their names consist of some reference information and additionally a type marker (CUR = currency, PRC = percentage, INT = number (including decimals!)). The last column contains the target value, the monthly premium. Not all columns are completely filled, which provides some additional challenge! About us **The role of comparis.ch on the insurance market** On 4 December 1994, the Swiss voters decided on a reform of the Health Insurance Act (KVG). This stipulated that policyholders should be able to choose and change their insurance companies freely. Therefore, since 1996, the health insurance companies have been in a competitive situation. comparis.ch has an important function in this competition: 1. Since 1996, the comparis.ch comparison has been making the complex premium jungle transparent, clear and comparable. This is the only way to guarantee that the policyholders are able to properly exercise their freedom of choice. 2. Transparency and comparability puts the health insurance companies under a greater pressure to offer cheaper premiums and good service quality. This also forces them to organise their administration in a particularly efficient manner, for example. 3. The IT processes at comparis.ch are very cost efficient. Calculations show that, due to the automated Comparis processes alone, the health insurance companies can save up to 500 million francs in administration costs. 4. comparis.ch points out the financial advantages that come with alternative insurance models and optional deductible rates. Thereby, these models are promoted, which is entirely in the spirit of the KVG. They cause the policyholders to take up more personal responsibility with regard to health care and curb the rise in health care costs. **About comparis** [https://en.comparis.ch/comparis/info/wir][2] comparis.ch is the leading Swiss Internet comparison service. On www.comparis.ch, consumers can easily and quickly compare rates and services of health insurance providers, other insurers, banks and telecommunications providers as well as offers for properties, cars and motorcycles. Thanks to the comparisons and ratings from comparis.ch, consumers can directly switch to the provider with the best price-performance ratio. comparis.ch was founded in 1996 by Richard Eisler, economist. Since June of 2000, comparis.ch is a public limited company based in Zurich and employs over 200 persons. comparis.ch also possesses an advisory board. Every quarter, comparis.ch publishes the Konsumentenstimme (in German and French only) with facts, figures and background information on current consumer issues. [1]: https://www.bag.admin.ch/bag/en/home.html [2]: https://en.comparis.ch/comparis/info/wir