Trouble using Quantity to extend existing DataFrame

[MichaelTiemannOSC]

This may be another manifestation of #26

I want to use a quantity to initialize a dataframe using a .loc indexer. I can use a floating point number in the expected way, but not a Quantity. What is the work-around?

import pandas as pd
import numpy as	np
import pint_pandas

from pint import UnitRegistry
ureg = UnitRegistry()
Q_ = ureg.Quantity

ureg.define("Fe_ton = [produced_ton]")
ureg.define("CO2 = [emissions]")

hist_dict = {2009: {('US6293775085', 'Productions', 'Production'): np.nan, ('US6293775085', 'Emissions', 'S1'): Q_(np.nan, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(np.nan, 'CO2 * metric_ton')},
 2010: {('US6293775085', 'Productions', 'Production'): np.nan, ('US6293775085', 'Emissions', 'S1'): Q_(1.65226001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(19599001.4, 'CO2 * metric_ton')},
 2011: {('US6293775085', 'Productions', 'Production'): np.nan, ('US6293775085', 'Emissions', 'S1'): Q_(1.62028001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(17902001.4, 'CO2 * metric_ton')},
 2012: {('US6293775085', 'Productions', 'Production'): np.nan, ('US6293775085', 'Emissions', 'S1'): Q_(1.58192001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(17256001.4, 'CO2 * metric_ton')},
 2013: {('US6293775085', 'Productions', 'Production'): np.nan, ('US6293775085', 'Emissions', 'S1'): Q_(1.69000001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(21000001.4, 'CO2 * metric_ton')},
 2014: {('US6293775085', 'Productions', 'Production'): Q_(91200001.4, 'Fe_ton'), ('US6293775085', 'Emissions', 'S1'): Q_(1.74000001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(17000001.4, 'CO2 * metric_ton')},
 2015: {('US6293775085', 'Productions', 'Production'): Q_(92479001.4, 'Fe_ton'), ('US6293775085', 'Emissions', 'S1'): Q_(1.76000001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(16000001.4, 'CO2 * metric_ton')},
 2016: {('US6293775085', 'Productions', 'Production'): Q_(90800001.4, 'Fe_ton'), ('US6293775085', 'Emissions', 'S1'): Q_(1.76000001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(14000001.4, 'CO2 * metric_ton')},
 2017: {('US6293775085', 'Productions', 'Production'): Q_(93100001.4, 'Fe_ton'), ('US6293775085', 'Emissions', 'S1'): Q_(1.79700001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(15100001.4, 'CO2 * metric_ton')},
 2018: {('US6293775085', 'Productions', 'Production'): Q_(92500001.4, 'Fe_ton'), ('US6293775085', 'Emissions', 'S1'): Q_(1.74900001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(13900001.4, 'CO2 * metric_ton')},
 2019: {('US6293775085', 'Productions', 'Production'): Q_(89800001.4, 'Fe_ton'), ('US6293775085', 'Emissions', 'S1'): Q_(1.69800001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(12100001.4, 'CO2 * metric_ton')},
 2020: {('US6293775085', 'Productions', 'Production'): Q_(71500001.4, 'Fe_ton'), ('US6293775085', 'Emissions', 'S1'): Q_(1.41300001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(9500001.4, 'CO2 * metric_ton')},
 2021: {('US6293775085', 'Productions', 'Production'): Q_(71500001.4, 'Fe_ton'), ('US6293775085', 'Emissions', 'S1'): Q_(1.41300001e+08, 'CO2 * metric_ton'), ('US6293775085', 'Emissions', 'S2'): Q_(9500001.4, 'CO2 * metric_ton')}}

hist_data = pd.DataFrame(data=hist_dict)
ei_keys = {'S1': ('US6293775085', 'Emission Intensities', 'S1'),
           'S2': ('US6293775085', 'Emission Intensities', 'S2'),
           'S3': ('US6293775085', 'Emission Intensities', 'S3'),
           'S1S2': ('US6293775085', 'Emission Intensities', 'S1S2'),
           'S1S2S3': ('US6293775085', 'Emission Intensities', 'S1S2S3')}
scope = 'S1'
production_units='Fe_ton'

try:
    hist_data.loc[ei_keys[scope], 2014] = Q_(np.nan, 't CO2') / Q_(np.nan, production_units)
    # *** TypeError: object of type 'float' has no len()                                                                                                                                                                                                 
except:
    print("*** TypeError: object of type 'float' has no len()")

hist_data.loc[ei_keys[scope], 2009] = 1.

print(hist_data)                                                                                                                                                                                                                        
# US6293775085 Productions          Production                   NaN                           NaN                           NaN  ...             89800001.4 Fe_ton             71500001.4 Fe_ton             71500001.4 Fe_ton
#              Emissions            S1          nan CO2 * metric_ton  165226001.0 CO2 * metric_ton  162028001.0 CO2 * metric_ton  ...  169800001.0 CO2 * metric_ton  141300001.0 CO2 * metric_ton  141300001.0 CO2 * metric_ton
#                                   S2          nan CO2 * metric_ton   19599001.4 CO2 * metric_ton   17902001.4 CO2 * metric_ton  ...   12100001.4 CO2 * metric_ton    9500001.4 CO2 * metric_ton    9500001.4 CO2 * metric_ton
#              Emission Intensities S1                           1.0                           NaN                           NaN  ...                           NaN                           NaN                           NaN
#
# [4 rows x 13 columns]

hist_data.loc[ei_keys[scope]] = 1.

print(hist_data)               
#                                                               2009                          2010                          2011  ...                          2019                          2020                          2021
# US6293775085 Productions          Production                   NaN                           NaN                           NaN  ...             89800001.4 Fe_ton             71500001.4 Fe_ton             71500001.4 Fe_ton
#              Emissions            S1          nan CO2 * metric_ton  165226001.0 CO2 * metric_ton  162028001.0 CO2 * metric_ton  ...  169800001.0 CO2 * metric_ton  141300001.0 CO2 * metric_ton  141300001.0 CO2 * metric_ton
#                                   S2          nan CO2 * metric_ton   19599001.4 CO2 * metric_ton   17902001.4 CO2 * metric_ton  ...   12100001.4 CO2 * metric_ton    9500001.4 CO2 * metric_ton    9500001.4 CO2 * metric_ton
#              Emission Intensities S1                           1.0                           1.0                           1.0  ...                           1.0                           1.0                           1.0
#
# [4 rows x 13 columns]                    

[andrewgsavage]

There is no workaround for initialising a column to contain a PintArray using df.loc. Use one of the ways listed at the bottom of the notebook.

[andrewgsavage]

the example now runs OK