search

yukiregista / DSC_STUDY

1 stars 3 forks source link

Geo Encoding #24

Open rkondo3 opened 5 months ago

Mimi33Mimi commented 4 months ago

I encoded City + State data into Geohash by packages geopy and geohash2. In both train and test data, I got missing data in more than 200 rows.

rkondo3 commented 4 months ago

Great Job! I tried the simplest model with your latitude and longitude cols. It improved the F1 score! The simplest model: 0.6283502461479223 The simplest model+lat+long: 0.6375352741316181

Mimi33Mimi commented 4 months ago

Many thanks! I submitted the errata in location into Google Drive.

The file errata_location.csv is a list of corrections for the geographical data I with latitudes, longitudes, and geohash. The names of locations that I couldn't convert before were due to two reasons: annotations in parentheses and typos.

I have successfully resolved these issues, and now I can convert them without any problems.

I apologize for not being familiar with using Pandas, and I couldn't rewrite my data I submitted. Please handle this matter.

ooyamakeigo commented 4 months ago

Thank you! It might seem simple, but I think it was more deceptively hard. I will use it!