[FIXED] Pandas dataframe: how to multiply a specific value by using its row index?

Issue

I want to change specific values in a Pandas dataframe. Here is an example dataframe (in reality, there are many more rows):

              Value                       Property
0               CH4                       Type
1          -10.90979                      Density (g/cm3)
2           5.00000                       Temperature (K)

Here I want to multiply "10.90979" by 10 in the row labeled "1". I don't want to have to write "10.90979 * 10" because I only know that I have a property called "Density (g/cm3)". I don't know its value. So I want to use the index of the row in which "Density (g/cm3)" appears in the multiplication.

I have tried:

row_index = df.index.get_loc(df[df['Property'] == 'Density (g/cm3)'].index[0])
new_value = df.iloc[row_index][0] * 10
df["Value"].replace(df.iloc[row_index][0], new_value, inplace=True)

However, this gives me weird output. I get:

              Value                       Property
0               CH4                       Type
1 -10.90979-10.90979...                   Density (g/cm3)
2           5.00000                       Temperature (K)

I can't post the details of the code but am hoping someone will recognize a simple mistake. I am not sure that I'm using multiplication correctly for a dataframe. I also tried using

df.iloc[row_index][0].mul(10)

but get the error AttributeError: 'str' object has no attribute 'mul'.

Can anyone please point me in the right direction?

Solution

As others pointed out, your Value column is not numerical (because of the 'CH4' first value). But, more fundamentally, why do you want to multiply the density by 10? What's the higher level problem?

From this question (and your previous one, now deleted), it looks like you have chemical compounds and are trying to do unit conversions on their physical properties. I would recommend looking into packages that do that readily. For example: pint and its pandas companion pint_pandas.

It seems that you have DataFrames with a chemical compound (e.g., 'SO2' or 'CH4') as the value of the first row, and then some physical properties.

You may be better off using dicts to represent your properties. E.g.:

from pint import UnitRegistry

Q_ = UnitRegistry().Quantity

compounds = {
    'CH4': dict(
        name='methane',
        molar_mass=Q_(16.043, 'g / mol'),
        density_0=Q_(0.657, 'kg/m**3'),
        at_0=Q_(25, '°C'),
        density_1=Q_(0.717, 'kg/m**3'),
        at_1=Q_(0, '°C'),
        density_2=Q_(422.8, 'g/L'),
        at_2=Q_(-162, '°C'),
    ),
    'SO2': dict(
        name='sulfur dioxide',
        molar_mass=Q_(64.066, 'g / mol'),
        density_0=Q_(2.6288, 'kg / m**3'),
        at_0=Q_(25, '°C'),
    ),
}

Then, if you like, you can convert this into a DataFrame:

import pandas as pd
import pint_pandas

df = pd.DataFrame(compounds).T

and convert a specific column into some desired unit, for example:

>>> df['density_0'].astype('pint[g/L]')
CH4    0.6570000000000001
SO2    2.6288000000000005
Name: density_0, dtype: pint[gram / liter]

# or (shudder):
>>> df['density_0'].astype('pint[lb/gal]')
CH4     0.0054829307249767
SO2    0.02193839922346841
Name: density_0, dtype: pint[pound / gallon]

or convert the whole DataFrame to have consistent units for each column:

z = df.astype(dict(
    molar_mass='pint[g/mol]',
    density_0='pint[g/L]',
    density_1='pint[g/L]',
    density_2='pint[g/L]',
    at_0='pint[degC]',
    at_1='pint[degC]',
    at_2='pint[degC]',
))

>>> z
               name  molar_mass           density_0  at_0           density_1  at_1  density_2  at_2
CH4         methane      16.043  0.6570000000000001    25  0.7170000000000001     0      422.8  -162
SO2  sulfur dioxide      64.066  2.6288000000000005    25                 nan   nan        nan   nan

>>> z.dtypes
name                        object
molar_mass       pint[gram / mole]
density_0       pint[gram / liter]
at_0          pint[degree_Celsius]
density_1       pint[gram / liter]
at_1          pint[degree_Celsius]
density_2       pint[gram / liter]
at_2          pint[degree_Celsius]
dtype: object

Answered By - Pierre D