Skip to content

data_structure_intro.md

Latest commit

 

History

History
163 lines (126 loc) · 6.57 KB

data_structure_intro.md

File metadata and controls

163 lines (126 loc) · 6.57 KB

Data Structure Introduction

The data is packaged as a .pkl file in the DataSet folder. You can find it in ./DataSet/ or download the dataset. The dataset file is over 200 MB in size. Due to GitHub's size limit, if you encounter download issues, please contact the author via email.

Data Dictionary

Loading the Data

The overall data is saved in a single .pkl file, which provides everything necessary for reproducing the results in Table 2 (Section 6). You can load the data usingß:

import BuildModels as bm 
import ToolFunctions as tf
import CompareFunc as cm

filename = 'DataSet/agg_dataOptionDictsparseVIPLuce_repeat36_2024-09-22-04-20-19.pkl'
dataOptionDict_repeat, probSettingSet, repeatNum, modelReport = tf.load(filename)

Data Components

The loaded data contains four parts:

  1. dataOptionDict_repeat: Provides all the necessary entries characterizing the instances.

  2. probSettingSet: Provides a pandas MultiIndex characterizing the problem size and type. Each index corresponds to a combination of problem parameters:

    (n, m), (alpha0), (v0_off, v0_on), (luce), (kappa_on, kappa_off), (knapsack_off, knapsack_on)

    The structure looks like:

    MultiIndex([( (100, 50), 0.5,  (1, 2), 0, (1, 1), (0, 0)),
            ( (100, 50), 0.5,  (1, 5), 0, (1, 1), (0, 0)),
            ( (100, 50), 0.5, (1, 10), 0, (1, 1), (0, 0)),
            ( (100, 50), 0.5,  (1, 2), 1, (1, 1), (0, 0)),
            ( (100, 50), 0.5,  (1, 5), 1, (1, 1), (0, 0)),
            ( (100, 50), 0.5, (1, 10), 1, (1, 1), (0, 0)),
            ( (150, 75), 0.5,  (1, 2), 0, (1, 1), (0, 0)),
            ( (150, 75), 0.5,  (1, 5), 0, (1, 1), (0, 0)),
            ( (150, 75), 0.5, (1, 10), 0, (1, 1), (0, 0)),
            ( (150, 75), 0.5,  (1, 2), 1, (1, 1), (0, 0)),
            ( (150, 75), 0.5,  (1, 5), 1, (1, 1), (0, 0)),
            ( (150, 75), 0.5, (1, 10), 1, (1, 1), (0, 0)),
            ((200, 100), 0.5,  (1, 2), 0, (1, 1), (0, 0)),
            ((200, 100), 0.5,  (1, 5), 0, (1, 1), (0, 0)),
            ((200, 100), 0.5, (1, 10), 0, (1, 1), (0, 0)),
            ((200, 100), 0.5,  (1, 2), 1, (1, 1), (0, 0)),
            ((200, 100), 0.5,  (1, 5), 1, (1, 1), (0, 0)),
            ((200, 100), 0.5, (1, 10), 1, (1, 1), (0, 0))]
)

    Parameter Definitions:

    • m: Number of products
    • n: Number of customer segments
    • alpha0: Arrival ratio of the offline customer segment
    • v0_off and v0_on: Non-purchase option utility for the offline and online segments, respectively
    • luce: Indicates whether it is subject to the 2-Stage-Luce-Model for the online channel (luce=1) or not (luce=0)
    • kappa_on and kappa_off: Right-hand side of the capacity constraints for online and offline segments, respectively
    • knapsack_off and knapsack_on: Right-hand side of the knapsack constraints for offline and online segments, respectively (coefficients should be provided separately)

  3. repeatNum: The total number of repetitions. This should be 36, as each combination was repeated 36 times.

  4. modelReport: The models to be compared. The expected value is:

    modelReport = ['Conic_Conic-mo', 'MC_Conv-mo-soc-aC', 'MILP']

    Where:

    • Conic_Conic-mo: The Conic formulation
    • MILP: The MILP formulation
    • MC_Conv-mo-soc-aC: The implementation of our proposed approach, corresponding to CH-K (see Algorithm 2 in the paper for details)

Understanding dataOptionDict_repeat

The dataOptionDict_repeat component plays the most important role in delivering the instance parameters, including:

  • Instance scale (number of customer segments and products)
  • Preference weight for each product for each customer segment
  • Revenue for each product for each customer segment
  • Online and offline customer arrival ratios
  • Adjacency matrix characterizing the dominating relationships used for the 2-Stage-Luce-Model

Structure of dataOptionDict_repeat

The dataOptionDict_repeat is a dictionary with keys range(0, repeatNum).

To view the keys:

>>> dataOptionDict_repeat.keys()

You will see:

dict_keys([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35])

Access a specific repetition:

>>> dataOptionDict = dataOptionDict_repeat[r]  # r is the repetition number

Structure of dataOptionDict

The dataOptionDict is a dictionary with keys from probSettingSet.

To see the keys:

>>> list(dataOptionDict.keys())

You will get:

    [((100, 50), 0.5, (1, 2), 0, (1, 1), (0, 0)),
    ((100, 50), 0.5, (1, 5), 0, (1, 1), (0, 0)),
    ((100, 50), 0.5, (1, 10), 0, (1, 1), (0, 0)),
    ((100, 50), 0.5, (1, 2), 1, (1, 1), (0, 0)),
    ((100, 50), 0.5, (1, 5), 1, (1, 1), (0, 0)),
    ((100, 50), 0.5, (1, 10), 1, (1, 1), (0, 0)),
    ((150, 75), 0.5, (1, 2), 0, (1, 1), (0, 0)),
    ((150, 75), 0.5, (1, 5), 0, (1, 1), (0, 0)),
    ((150, 75), 0.5, (1, 10), 0, (1, 1), (0, 0)),
    ((150, 75), 0.5, (1, 2), 1, (1, 1), (0, 0)),
    ((150, 75), 0.5, (1, 5), 1, (1, 1), (0, 0)),
    ((150, 75), 0.5, (1, 10), 1, (1, 1), (0, 0)),
    ((200, 100), 0.5, (1, 2), 0, (1, 1), (0, 0)),
    ((200, 100), 0.5, (1, 5), 0, (1, 1), (0, 0)),
    ((200, 100), 0.5, (1, 10), 0, (1, 1), (0, 0)),
    ((200, 100), 0.5, (1, 2), 1, (1, 1), (0, 0)),
    ((200, 100), 0.5, (1, 5), 1, (1, 1), (0, 0)),
    ((200, 100), 0.5, (1, 10), 1, (1, 1), (0, 0))]

Access a specific problem setting:

probName, (data, option) = dataOptionDict[probSetting_info]  
# probSetting_info is a key (index from probSettingSet) that locates the specific problem

Components:

  • probName: The name of this problem instance
  • (data, option): Contains the parameters including:
    • Preference weights
    • Revenue
    • Arrival ratios
    • 2-Stage-Luce-Model characterizing parameters
    • Additional problem-specific data

These use custom classes defined in BuildModels.py. See the Option and Data class definitions for details. These objects are the necessary inputs for the subsequent solving procedures.

Converting Data to Text Format

If you want to convert the data to .txt files, use the provided data_save_to_txt.py script. This is also helpful for understanding the data structure.

Note

The datasets for other numerical results follow a similar structure. See readme.md for details.