Add dataset creation utilities

dataset-creation-utilities/create-year-lists.py

@@ -0,0 +1,13 @@
+import pandas as pd
+import numpy as np
+from tqdm import tqdm
+
+df = pd.read_feather("hupd_metadata_2022-02-22.feather", columns=['application_number', 'filing_date'])
+assert df['filing_date'].dt.year.isnull().sum() == 2
+df['year'] = df['filing_date'].dt.year.fillna(2018.0).astype(int)
+years = df['year'].value_counts()
+print(years)
+for year in tqdm(years.index):
+    df_year_numbers = df['application_number'][df['year'] == year]
+    df_year_numbers.to_frame().to_csv(f'lists/{year}.txt', sep=' ', index=False, header=False)
+    print(f'saved {year}')

dataset-creation-utilities/move-files.sh

@@ -0,0 +1,5 @@
+for year in 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018  # 2004 
+do
+	mkdir years/${year}
+	cat hupd-subset/lists/$year.txt | xargs -i mv distilled/{}.json years/${year}
+done