Combination of values for sample selection

Question

Guys, I’m having difficulties to finish a job and would like help. Anyway, below is the excerpt from my list.csv:

n,NOME,COR,VALOR
1,Lapis ,Verde,"11,1"
2,Caneta ,Vermelho,"12,25"
3,Lapiseira ,Azul,"15,45"
...

and the code I wrote in Python:

from itertools import combinations
import csv

def get_column_of_csv(filename, column):
    with open(filename) as stream:
        catalogo = csv.DictReader(stream)
        for row in catalogo:
            yield row[column]

for name in get_column_of_csv('lista.csv', 'VALOR'):
    for i in (2, len(name)):
            for comb in combinations(name, i):
                if sum(int(comb)) <= 25:
                    print (comb,'<= 25,00')

    print(name)

My intention is to make the code combine values so that their sum does not exceed certain established value and at the end show me the names of the items whose pairs did not exceed this value. An example of output would be: "Pencil + Pencil" (Since the sum of these does not exceed the established value of 25,00).

I saw this post: LINK based on my code, but I was unsuccessful.

ERROR:

Traceback (most recent call last):
  File "kits.py", line 13, in <module>
    if sum(int(comb)) <= 25:
TypeError: int() argument must be a string, a bytes-like object or a number, not 'tuple'

If anyone can help me... I’d be grateful.

EXTRA: If in the list of items I had the name of the item together with the manufacturer (ex: BIC Pen, Faber-Castell Pen,...) and I wanted to assemble only combination of items of a certain brand, or exclude a certain brand from the filter, as could be done?

Answer 1

The error is basically because you are trying to convert a tuple into a whole. Basically doing something like:

int((2, 5, 7))

And the class int don’t know what to do with a tuple.

I found some points of code a little confusing, mainly the line for i in (2, len(name)); did not understand why you will generate combinations with 2 items and with len(name) items. How you are accessing the column 'VALOR', the value of name will be, for example, "11,1", and so len(name) would be 4. After that you generate the combinations in name with i items, but this would generate combinations between the characters of "11,1" that is not what you intend to do.

Your code will be much simpler if you separate the responsibilities by functions. The first step is to get the file data, then we do the function that reads the file returning a generator to the lines already converting the value of VALOR for float:

def read_csv(path):
    with open(path) as stream:
        reader = csv.DictReader(stream)
        for row in reader:
            row['VALOR'] = float(row['VALOR'].replace(',', '.'))
            yield row

From the result we want to create the combinations 2 to 2, so we use the function itertools.combinations:

data = read_csv('data.csv')
combinations = itertools.combinations(data, 2)

Of all combinations only those that have the column sum VALOR less than or equal to 25. For this we create a function that will receive the sequence of combinations, the name of the column we want to add and the desired limit:

def sum_of_column_le(combinations, column, limit):
    for combination in combinations:
        sum_of_column = sum(product.get(column, 0) for product in combination)
        if sum_of_column <= limit:
            yield combination

So, to get the combinations that have the sum of VALOR less than or equal to 25 do:

result = sum_of_column_le(combinations, 'VALOR', 25.00)

And so we can iterate on result to display the combinations that interest us:

for combination in result:
    print([product['NOME'] for product in combination])

For the question data:

n,NOME,COR,VALOR
1,Lapis ,Verde,"11,1"
2,Caneta ,Vermelho,"12,25"
3,Lapiseira ,Azul,"15,45"

The exit would be:

['Lapis ', 'Caneta ']

For it is the last combination of products that has the sum of values less than or equal to 25.

^{See working in Repl.it | Github GIST}

Answer 2

TL;DR

A solution with pandas, I used the data of your example but added the brands to demonstrate the filtering, eliminating the lines with the brand "bar":

import io
import pandas as pd
from itertools import combinations

# Simulando o csv
csv = '''
nome,cor,marca,valor
Lapis, Verde, bic,"11,1"
Caneta, Vermelho, cristal,"12,25"
Lapiseira, Azul, foo,"15,45"
Lapis, Verde, foo,"12,00"
borracha, Branca, bar,"10,00"
borracha, Verde, bar,"11,00"
'''

# Lendo o csv 
df = pd.read_csv(io.StringIO(csv))

# Filtrando o df para que nao contenha a marca 'bar'
df = df[~df.marca.str.contains('bar')]

# Criando coluna para uma tupla de descricao e valor (adapte ao seu contexto)
df['descricao'] = df.nome + df.cor + df.marca
df['item'] = list(zip(df.descricao, df.valor))

# Apresentando o dataframe resultante
print(df)

Exit:

        nome        cor     marca  valor                descricao  \
0      Lapis      Verde       bic   11.1          Lapis Verde bic   
1     Caneta   Vermelho   cristal  12.25  Caneta Vermelho cristal   
2  Lapiseira       Azul       foo  15.45       Lapiseira Azul foo   
3      Lapis      Verde       foo  12.00          Lapis Verde foo   

                               item  
0           (Lapis Verde bic, 11.1)  
1  (Caneta Vermelho cristal, 12.25)  
2       (Lapiseira Azul foo, 15.45)  
3          (Lapis Verde foo, 12.00)  

The next step is to combine the items into a list:

# Fazendo as combinacoes
combs = list(combinations(df.item, 2))

# Apresentando as combinacoes
print(combs)

Exit:

[(('Lapis Verde bic', '11.1'), ('Caneta Vermelho cristal', '12.25')),
 (('Lapis Verde bic', '11.1'), ('Lapiseira Azul foo', '15.45')),
 (('Lapis Verde bic', '11.1'), ('Lapis Verde foo', '12.00')),
 (('Caneta Vermelho cristal', '12.25'), ('Lapiseira Azul foo', '15.45')),
 (('Caneta Vermelho cristal', '12.25'), ('Lapis Verde foo', '12.00')),
 (('Lapiseira Azul foo', '15.45'), ('Lapis Verde foo', '12.00'))]

Finally we go through the list of combinations, selecting the ones that meet the condition, in the sum of the values <= 25

# Selecionando as combinacoes validas
valid_combs = []
max_value = 25
for comb in combs:
    if float(comb[0][1]) + float(comb[1][1]) <= max_value:
        valid_combs.append(comb)

# Apresentando o resultado
print(valid_combs)        

Output (Tuples with combinations <= 25):

[(('Lapis Verde bic', '11.1'), ('Caneta Vermelho cristal', '12.25')),
 (('Lapis Verde bic', '11.1'), ('Lapis Verde foo', '12.00')),
 (('Caneta Vermelho cristal', '12.25'), ('Lapis Verde foo', '12.00'))]

Putting the final result into one pandas.DataFrame:

final_list = []
for valid in valid_combs:
    final_list.append((valid[0][0],valid[0][1], valid[1][0],valid[1][1], 
                       float(valid[0][1])+float(valid[1][1])))

df_final = pd.DataFrame(final_list, columns=['Item 1', 'Valor', 
                                              'Item 2', 'Valor', 'Total'])

# Apresentando o resultado final
print(df_final)

Exit:

                    Item 1  Valor                   Item 2  Valor  Total
0          Lapis Verde bic   11.1  Caneta Vermelho cristal  12.25  23.35
1          Lapis Verde bic   11.1          Lapis Verde foo  12.00  23.10
2  Caneta Vermelho cristal  12.25          Lapis Verde foo  12.00  24.25

Filtering the DataFrame final:

Let’s filter the final result by removing the lines in which the word "Pen"

# Filtrando o df, removendo as combinacoes q tenham "Caneta"
df_filtered = df_final[~df_final['Item 1'].str.contains('Caneta') & 
                       ~df_final['Item 2'].str.contains('Caneta')]

# Apresentando o dataframe filtrado
print('','DataFrame filtrado: ', df_filtered,sep='\n')

Exit

DataFrame filtrado: 
            Item 1 Valor           Item 2  Valor  Total
1  Lapis Verde bic  11.1  Lapis Verde foo  12.00   23.1

See working on repl.it.

Answer 3

I changed your code a little bit and it worked right here.

What I did:

• I took the column NAME and played all the items on a list;
• I took the column WORTH and played all the items on a list;
• I made a replace in the columnar WORTH and changed all the ',' for '.' to convert to float.
• I discovered the index where the sum of products is below 25.

With that, the code went like this:

from itertools import combinations
import csv

lista_valor = []
lista_produto = []
combinacoes_possiveis = []
resultado = []

def get_column_of_csv(filename, column):
    with open(filename) as stream:
        catalogo = csv.DictReader(stream)
        for row in catalogo:
            yield row[column]

for produto in get_column_of_csv('lista.csv', 'NOME'):

    lista_produto.append(produto)


for name in get_column_of_csv('lista.csv', 'VALOR'):
    name = str(name)                # TRANSFORMA EM STRING PARA PODER USAR REPLACE
    name = name.replace(',', '.')   # TROCA , POR . PARA PODER CONVERTER EM FLOAT
    name = float(name)              # CONVERTE EM FLOAT

    lista_valor.append(name)        # ADICIONA O VALOR A UMA LISTA


# --------BUSCA COMBINAÇÃO DE 2 VALORES DA LISTA-----------------
# --------------QUE NÃO ULTRAPASSAM 25---------------------------

for t in combinations(lista_valor,2):
    if sum(t) <= 25:
        idx = [lista_valor.index(a) for a in t]    # PEGA O INDEX NO QUAL A COMBINAÇÃO
                                                   # NÃO ULTRAPASSA 25

        combinacoes_possiveis.append(idx)



print(combinacoes_possiveis)     # PARES DE ÍNDICES POSSÍVEIS

for i in range(len(combinacoes_possiveis)):  ### PREVINE RESULTADOS REPETIDOS

  if str(lista_produto[combinacoes_possiveis[i][0]]) + 'e ' + 
  str(lista_produto[combinacoes_possiveis[i][1]]) in resultado:
    print('Elemento repetido')

  else:

    resultado.append(str(lista_produto[combinacoes_possiveis[i][0]]) + 'e ' +
    str(lista_produto[combinacoes_possiveis[i][1]]))


print(resultado)     # ['Lapis e Caneta ', 'Lapis e Lapiseira ', 'Lapis e Boracha ', 'Lapis e Caneta Colorida ', 'Lapis e Corretivo ', 'Caneta e Lapiseira ', 'Caneta e Boracha ', 'Caneta e Caneta Colorida ', 'Caneta e Corretivo ', 'Lapiseira e Boracha ', 'Lapiseira e Lapiseira ', 'Lapiseira e Caneta Colorida ', 'Lapiseira e Corretivo ', 'Boracha e Lapiseira ', 'Boracha e Caneta Colorida ', 'Boracha e Corretivo ', 'Caneta Colorida e Corretivo ']

Good luck with your studies!

EDIT:

I changed all the CSV values so there were 21 possible combinations.