How to assign multiple values to a key using dictionaries?

Question

I have a csv file that contains the name (of a video game), platform, Genre, Publisher, etc. I am trying to create 3 separate dictionaries. Dictionary one was easy since the key used was the title of a video game which is unique.

For the 2nd and 3rd dictionary, I am having issues since the keys "Genre" and "Publisher" are not unique. I am trying to have D2 look like:

D2 = { 'Puzzle' : [(tup2),(tup2], 'Another genre': [(tup2)]...}

Since there are multiple games that have the same genre.

import csv

fp = open("video_game_sales_tiny.csv", 'r')
fp.readline()
reader = csv.reader(fp)

D1 = {}
D2 = {}
D3 = {}

for line in reader:
    name = line[0].lower().strip()
    platform = line[1].lower().strip()

    if line[2] in (None, 'N/A'):
        pass
    else:
        year = int(line[2])

    genre = line[3].lower().strip()
    publisher = line[4]
    na_sales = float(line[5])
    europe_sales = float(line[6])*1000000
    japan_sales = float(line[7])*1000000
    other_sales = float(line[8])*1000000

    global_sales = (europe_sales + japan_sales + other_sales)
    tup = (name,platform, year,genre, publisher, global_sales)
    tup2 = (genre, year, na_sales, europe_sales, japan_sales, other_sales, global_sales)
    tup3 = (publisher, name, year, na_sales, europe_sales, japan_sales, other_sales, global_sales)

    D1[name] = tup
    D2[genre] = tup2
    D3[publisher] = tup3

print(D1)
print(D2)
print(D3)

Answer 1

You should create the entry for genre (for instance) as a list, and then append to the list.

 if not genre in D2:
      D2[genre] = []

 D2[genre].append(tup2)

Answer 2

You have a problem with non-unique keys.

If that problem is corrected (you need unique keys), the merge( ) method can be used with any other how options (left, right, inner, ...).

The Pandas Library merge() method is very powerful and will solve your problem.

But, you need to do something about non-unique keys problem.

I suggest use the method unique() and make your own list of indexes for each DataFrame . This will be only one more layer in your ETL process.

Suppose you have two DataFrames : df_a and df_b . This dataframes share a unique key called u_key .

The merge process with these dataframes will be something like:

import pandas as pd
...
left_merge = pd.merge(df_a, df_b, on=["u_key"], how="left")