While working on a project that involved looking at King County housing data and working on creating a model to predict pricing, I was intrigued during the data cleaning process. Since I live in King County, I was interested in the data because I see the housing costs around me. When examining the data, I found that the column indicating whether a property was waterfront property or not had many Nan entries.
I tried a couple different approaches to figure out how to deal with these unknown values. Knowing more about King County, when I checked to see how many properties were listed as waterfront, it seemed a bit low given how much water exists in this county.
The first was to look at the houses on a map and see if it was easy to designate waterfront properties based on their latitude and longitude. I used geomapping to display both the waterfront and non-waterfront properites. The red is for non-waterfront properties, the blue is for waterfront, and the yellow has waterfront listed as a Nan entry.
As you can see, the yellow is very well dispersed amongst the red and blue. Along with that, it is difficult to determine yellow areas that should be blue, because there are also red houses near them. I then looked to see how much difference in cost there was between waterfront properties and non-waterfront properties and determined if I changed the Nan entries to the mode (0), it wouldn’t change our data too drastically.
However, I did enjoy the process of learning how to plot the data geographically. I know I could extend it by adding in the actual King County map, but am still learning how to do that. I enjoyed the fact that the data looks like King County as I know it and I could easily identify the various bodies of water that exist in the county.