Import NSF AST funding data in CSV format

import_nsf_ast_csv.py

import pandas
import numpy as np
awards = pandas.read_csv("awards_dump_2014-06-20.csv",
                         parse_dates=['StartDate', 'LastAmendmentDate', 'ExpirationDate', 
                                      'AwardedAmountToDate'],
                         converters={'AwardedAmountToDate': lambda x: 
                                     float(x.replace('$', '').replace(',', ''))},
                         dtype={'AwardedAmountToDate':np.float64})
awards['DurationYears'] = ((awards['ExpirationDate'] - awards['StartDate']) / 
                           (365.25 * np.timedelta64(1, 'D')))
# unsure why, but AwardedAmountToDate wasn't correctly converted to float
awards['AwardedAmountToDate'] = awards['AwardedAmountToDate'].astype(float)
# A few awards have ExpirationDate < StartDate or implausibly long durations.  
# Let's eliminate those. (That's 17 out of ~6500)
awards = awards[(awards['DurationYears'] > 0) & (awards['DurationYears'] <= 11.)]