Pandas to_dict unwantedly modifying float numbers

Question

My code below takes in CSV data and uses pandas to_dict() function as one step in converting the data to JSON. The problem is it is modifying the float numbers (e.g. 1.6 becomes 1.6000000000000001). I am not concerned about the loss of accuracy, but because users will see the change in the numbers, it looks amateurish.

I am aware:

• this is something that has come up before here, but it was two years ago, was not really answered in a great way,
• also I have an additional complication: the data frames I am looking to convert to dictionaries could be any combination of datatypes

As such the issue with the previous solutions are:

1. Converting all the numbers to objects only works if you don't need to (numerically) use the numbers. I want the option to calculate sums and averages which reintroduces the addition decimal issue.
2. Force rounding of numbers to x decimals will either reduce accuracy or add additional unnecessary 0s depending on the data the user provides

My question:

Is there a better way to ensure the numbers are not being modified, but are kept in a numeric datatype? Is it a question of changing how I import the CSV data in the first place? Surely there is a simple solution I am overlooking?

Here is a simple script that will reproduce this bug:

import pandas as pd

import sys
if sys.version_info[0] < 3:
    from StringIO import StringIO
else:
    from io import StringIO

CSV_Data = "Index,Column_1,Column_2,Column_3,Column_4,Column_5,Column_6,Column_7,Column_8\nindex_1,1.1,1.2,1.3,1.4,1.5,1.6,1.7,1.8\nindex_2,2.1,2.2,2.3,2.4,2.5,2.6,2.7,2.8\nindex_3,3.1,3.2,3.3,3.4,3.5,3.6,3.7,3.8\nindex_4,4.1,4.2,4.3,4.4,4.5,4.6,4.7,4.8"

input_data = StringIO(CSV_Data)
df = pd.DataFrame.from_csv(path = input_data, header = 0, sep=',', index_col=0, encoding='utf-8')
print(df.to_dict(orient = 'records'))

Answer 1

I need to make df.to_dict('list') with right float numbers. But df.to_json() doesn't support orient='list' yet. So I do following:

 list_oriented_dict = {
    column: list(data.values())
    for column, data in json.loads(df.to_json()).items()
}

Not the best way, but it works for me. Maybe some one has a more elegant solution?