ruslansco / foursquare-data-analysis

Exploring popular venues in New York City & Tokyo, Modeling datasets and predicting users behavior using machine learning algorithms.

Analysis of Fousquare Check-Ins in New York City and Tokyo.

1. Preparation of the Datasets (dataset_TSMC2014_NYC.csv, dataset_TSMC2014_TKY.csv)

2. Converted string type values of column - “utcTimestamp” to datetime64[ns] type and disassembled into columns: “Year”, ”Month”, ”Hour”, ”Minute”.

3. Mapped latitude and longitude with precision 5, and inserted into new column - “Geohash”, by using pygeohash module. It divided the coordinates into “buckets” of different zones based on number of digits (precision).

4. Encoding data types Last step of data wrangling before I begin the model evaluation, was encoding Binary ID’s and string values to integer type, by using Label Encoder module. Specifically target value – Venue Category and feature – Geohash.

5. Target Value: Venue Category. Features: Year, Month, Weekday, Hour, Minute, and Geohash.

6. **Datasets Validations **

Files

foursquare_checkins_scatters.py

popular_venues_charts.py

evaluate_best_model.py