With the socio-economic development, most parts of the world have been or are witnessing a huge wave of urbanization. As a result, most of the people from remote rural areas move to big cities for the opportunity of the development which not only brings trouble to the management of the city, but also brings some trouble to themselves. Because they are not only get money in the process, they have to spend a lot of money for their basic daily life, such as food cost, clothes cost, and house rent accounting for most therein. It is indeed a big burden for those who have little income. In order to control the rent price in a rational range, many cities have developed a corresponding rent standard. But what factors will affect the price of rent. Through common sense, we may think there are some potential influencing factors, such as, household size, year of construction, equipment, the number of rooms, living environment and so on. But which is the main influencing factor? And how do these factors affect the rent price? These questions will be further explored in this paper.
The dataset used in this paper is derived from the kknn package in the R language environment. You can load the data by importing the code “data(“miete”)”. The discription of the data is showed below. A lot of German cities set up rent standards to provide the instructions to landlords, tenants and others. The standards are employed specially for the determination of the local rent (i.e. net rent as a function of household size, room equipment, etc.). To compose the rent standards, a random sample is conducted from all related households.The dataset consist of the data of 1082 households surveyed to compose the munich rent standard . This data set contains the 17 variables: nm, wfl, bj, bad0, zh, ww0, badkach, fenster, kueche, mvdauer, bjkat, wflkat, nmqm, rooms, nmkat, adr, wohn. Among these enormous variables, we choose the nm variable as the response variables. Then we choose all quantitative variables, that is, wfl,bj,mvdauer,rooms, and the first two binary variables, that is, bad0 and zh as the predictor variables. So the new miete contains 1082 rows and 7 columns,that is, 1082 observations and 7 variables. We know that the binary variables are transformed to dummy variables to fit the linear model, but that does not influence our analysis. The variables applied in this report is listed in details as followed.
For the variables used in the miete dataset, nm is the response variable, and the wfl, bj, bad0, zh, mvdauer and rooms are all chosen as the predictor variable. And for bad0 and mvdauer, the correlation coefficients are less than 0, which indicates the nm variance is opposite with these variables. The net rent increases 0.01 DM with the wfl increasing 1 sqm; the net rent increase 9.7*10-4 DM with the bj increasing 1 year. The net rent increases 0.16 DM for households with bathroom; The net rent increases 0.3 DM for households with central heating; the net rent decreases -0.009 DM with the mvdauer increasing 1 year; the net rent decreases -0.036 DM with the rooms increasing 1 room.