Key error-0

Answered

Milad K.E.

October 04, 2023 06:53 Edited

• 
• 
• 

I want to find an optimiser solution for financial resource allocation among 353 branches based on their performances.

Data are in a excel fime uploaded to Google drive. I am not sure after importing data to the data frame, I should define them one by one, or only the decision variable 'Fin' is enough.

Right now, I used the coefficient in objective function without defining them.

I received the 'KeyError: 0'. I would be grateful if you could share your opinion with me. Here is the code and the error text after that:

#Path to the Excel file in Google Drive

file_path = '/content/drive/My Drive/data.xlsx'

# Read data from the Excel file into a DataFrame

df = pd.read_excel(file_path) m = Model()

# I am not sure I should add all the variables from my table one by one here, or since the data frame loaded with my data, it is not necessary, I only added my decision variable 'Fin'

Fin = m.addVars(df['Fin'], name='x', vtype=GRB.CONTINUOUS)

# Number of fire departments

num_fire_departments = len(df) print("Number of Fire Departments:", num_fire_departments)

# Objective function, can I use exponencial (e^x) in the objective function?

objective_expr = quicksum( (df['Qi_PotenLoss'][i] * df['ai_VulnerCoeff'][i] * math.exp(df['Effici'][i] * Fin[i])) for i in range(num_fire_departments) ) m.setObjective(objective_expr, GRB.MINIMIZE)

 

System return:

Number of Fire Departments: 353

---------------------------------------------------------------------------

KeyError Traceback (most recent call last)

<ipython-input-7-a159bed2d57e> in <cell line: 11>()

      9 

     10 # Objective function

---> 11 objective_expr = quicksum(

     12 (df['Qi_PotenLoss'][i] * df['ai_VulnerCoeff'][i] * math.exp(df['Effici'][i] * Fin[i]))

     13 for i in range(num_fire_departments)

 

src/gurobipy/gurobi.pxi in gurobipy.quicksum()

 

<ipython-input-7-a159bed2d57e> in <genexpr>(.0)

     10 # Objective function

     11 objective_expr = quicksum(

---> 12 (df['Qi_PotenLoss'][i] * df['ai_VulnerCoeff'][i] * math.exp(df['Effici'][i] * Fin[i]))

     13 for i in range(num_fire_departments)

     14 )

 

KeyError: 0 

0

• 

  Riley Clement

  • Gurobi Staff

  October 04, 2023 23:25

  Hi Milad,

  If we look at the error message then you know that the error is in this line:

  df['Qi_PotenLoss'][i] * df['ai_VulnerCoeff'][i] * math.exp(df['Effici'][i] * Fin[i])) for i in range(num_fire_departments)

  and it is a KeyError, with key = 0 (not a Gurobi error).

  This means at least one of these values does not exist:

  df['Qi_PotenLoss'][0]
  df['ai_VulnerCoeff'][0]
  df['Effici'][0]
  Fin[0]

  It should be straightforward to discover which of these is causing an issue.

  - Riley

  0
• 

  Milad K.E.

  October 05, 2023 16:19

  • 
  • 
  • 

  Many thanks Raily!
  Do you think I should define all the variable the I define the 'Fin' to resolve it?

  0
• 

  Riley Clement

  • Gurobi Staff

  October 06, 2023 00:59

  Hi Milad,

  I'm afraid I have no intuition on how to resolve it without knowing which of the values caused an error, and why.

  - Riley

  0
• 

  Milad K.E.

  October 10, 2023 07:50 Edited

  • 
  • 
  • 

  Thank you very much Reily!

  
  I printed the values and all are imported and printed, the names are same as what I used in the objective function.

  Would you please take look at it?

  Also I need your advice on using Exp (exponencial) command in objective function.

  Here are the printed data:

   FDCode Fin Qi_PotenLoss qi_ActualLoss ai_VulnerCoeff Effici 
  0 FD101 245887.76 1.047386e+12 206115 2.141628e-07 0.34406 
  1 FD102 235504.29 5.361513e+11 5764620 1.170838e-05 0.36189 
  2 FD104 215294.51 6.495308e+11 2600340 4.355603e-06 0.39163 
  3 FD105 228721.26 8.127873e+11 2393325 3.160836e-06 0.30984 
  4 FD106 372830.97 1.099356e+12 8465055 8.469634e-06 0.25552
  
  Index(['FDCode', 'Fin', 'Qi_PotenLoss', 'qi_ActualLoss', 'ai_VulnerCoeff', 
  'Effici'],

  The obective function (which also has the keyerror-0) is:

  objective_expr = quicksum( (df['Qi_PotenLoss'][i] * df['ai_VulnerCoeff'][i] * math.exp(df['Effici'][i] * Fin[i])) for i in range(num_fire_departments) ) m.setObjective(objective_expr, GRB.MINIMIZE) 
  dtype='object')

  Apologies for the typing format I am using my mobile.

  Best, Milad

  0
• 

  Riley Clement

  • Gurobi Staff

  October 10, 2023 08:34

  Hi Milad,

  Let's address the key error and then we can look at the question of exponentials in the objective function.

  I can see what it going wrong here.  When you declare your Fin variables:

  Fin = m.addVars(df['Fin'], name='x', vtype=GRB.CONTINUOUS)

  the index of these variables is the "Fin" values in your dataframe, i.e. "FD101", "FD102", ...

  These are the keys in the Fin tupledict (our special Gurobi dictionary).  So Fin[0] will cause an error because 0 is not a value in the "Fin" column of your dataframe.  A workaround is to use Fin[df['Fin'][i]] instead of Fin[i], but now we are not being very clever with our formulation.

  We also will run into a problem where we cannot simply use a Gurobi object, such as a variable or linear expression, as an argument to math.exp function.  Could you please review our documentatino for the addGenConstrExp function, perhaps experiment with it on a simple dummy model, and then we can take it from there.

  Cheers,
  Riley

   

  0

Please sign in to leave a comment.