Python 通过嵌套字典排序

Question

From multiple tables I'm getting values of the form: (Sector, Stock, InvestedValue).

I'm using a Python dictionary object to insert these values and during insert if the combination (sector, stock) exists add the InvestedValue to the existing entry else add new entry to dictionary. After all data insert, let's say I end up with nested dictionary like:

stock_dict = {
    "Financial": {"HDFC Bank": 230.25, "Axis Bank": 70.15, "ICICI Bank": 110.45},
    "Automobile": {"Tata Motors": 135.67},
    "Consumer Goods": {"Avenue Supermarket": 190.45, "Godrej Industries": 120.32}
}

How to print through this nested dictionary in sorted fashion:

1. Get combination of Sector,Company,InvestedValue sorted on the basis of InvestedValue

2. Get combination of Sector, sum(InvestedValue) ie. sum of InvestedValue of each company in that sector again sorted on the sum

My current approach to solve these problems is to flatten the nested dictionary to list of tuples and run sorted on it. For example:

To solve 1:

stock_list = []
for sector in stock_dict:
    for stock in stock_dict[sector]:
        stock_list.append((sector, stock, stock_dict[sector][stock]))
sorted_list = sorted(stock_list, key=lambda stock: stock[2], reverse=True)

To solve 2:

sector_list = []
for sector in stock_dict:
    sector_list.append((sector,sum(stock_dict[sector].values())))
sorted_sector_list = sorted(sector_list, key=lambda sector: sector[1], reverse=True)

IS there a better approach ie sort directly on the nested dictionary without having to flatten it into the list?

Answer 1

If you uses pandas, it can convert you dict to a dataframe. You can then unstack the dataframe which will make it long form instead of wide. Once it's in the long format, you will have null values for any sector which a company does not participate in, you can drop those with dropna(), reset the index and sort values by your desired column. Once this is done rename your columns to the desired names, and select them in the order you want. You can take that same df and groupby sector and sum the InvestedValue.

import pandas as pd
stock_dict = {
    "Financial": {"HDFC Bank": 230.25, "Axis Bank": 70.15, "ICICI Bank": 110.45},
    "Automobile": {"Tata Motors": 135.67},
    "Consumer Goods": {"Avenue Supermarket": 190.45, "Godrej Industries": 120.32}
}

df = pd.DataFrame.from_dict(stock_dict, orient='index')
df = df.unstack().dropna().reset_index(name='InvestedValue').sort_values(by='InvestedValue', ascending=False)

df.columns = ['Company','Sector','InvestedValue']
df[['Sector','Company','InvestedValue']]

Output

           Sector             Company  InvestedValue
0       Financial           HDFC Bank         230.25
4  Consumer Goods  Avenue Supermarket         190.45
3      Automobile         Tata Motors         135.67
5  Consumer Goods   Godrej Industries         120.32
2       Financial          ICICI Bank         110.45
1       Financial           Axis Bank          70.15

2nd Part

df.groupby('Sector')['InvestedValue'].sum().reset_index().sort_values(by='InvestedValue', ascending=False)

Output

           Sector  InvestedValue
2       Financial         410.85
1  Consumer Goods         310.77
0      Automobile         135.67