VIC-on-QS-Rank

Machine learning project where we implement 10-fold cross-validation with 5 models:

• Logistic Regression
• Random Forest
• SVM
• LDA
• Naive Bayes

Getting Started

To deploy the code, run every cell of the vicQSrank.ipynb

Prerequisites

What things you need to install the software and how to install them

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import sklearn as sk

Installing

A step by step series of examples that tell you how to get a development env running

Run the following five method functions:

svmFunct(X,y)
randomForestFunct(X,y)
logregFunct(X,y)
ldaFunct(X,y)
naiveBayesFunct(X,y)

VIC Implementation cell 

And two dictionaries are stored after running this cells.

bestAveragesScores
bestMaxScore

The bestAveragesScores dictionary stores the 5 functions averages of the AUC per data base. bestMaxScore stores the name of the best model and the AUC of that model

Running the tests

Each classifier has its corresponding function, the VIC implementation calls the 5 function's model and stores the results of each model as a dictionary.

Break down into end to end tests

A run of one fold of every function returns the following values.

[[16  0]
 [ 4  1]]
AUC: 0.6
[[14  0]
 [ 4  2]]
AUC: 0.6666666666666666
[[11  2]
 [ 6  1]]
AUC: 0.49450549450549447
[[16  0]
 [ 2  2]]
AUC: 0.75
[[11  1]
 [ 7  1]]
AUC: 0.5208333333333334
[[14  0]
 [ 5  1]]
AUC: 0.5833333333333334
[[12  3]
 [ 1  4]]
AUC: 0.8
[[15  0]
 [ 4  1]]
AUC: 0.6
[[15  2]
 [ 2  1]]
AUC: 0.6078431372549019
[[12  4]
 [ 3  1]]
AUC: 0.5
[ 136   12 ]
[ 38   15 ]
10-FOLD AVERAGE OF SVM IS: 0.6009688934217237

And coding style tests

Three partitions of bestAveragesScores sample:

{'/content/drive/My Drive/datasets/dataset_100.csv': {'LDA': 0.760940594059406,
  'Logistic Regression': 0.8603960396039605,
  'MAX Score': ('Random Forest', 0.8704455445544554),
  'Naive Bayes': 0.8303465346534653,
  'Random Forest': 0.8704455445544554,
  'SVM': 0.8501485148514851},
 '/content/drive/My Drive/datasets/dataset_101.csv': {'LDA': 0.7714851485148515,
  'Logistic Regression': 0.855990099009901,
  'MAX Score': ('Random Forest', 0.8658910891089109),
  'Naive Bayes': 0.8263861386138613,
  'Random Forest': 0.8658910891089109,
  'SVM': 0.8463366336633663},
 '/content/drive/My Drive/datasets/dataset_102.csv': {'LDA': 0.7918894830659537,
  'Logistic Regression': 0.85650623885918,
  'MAX Score': ('Random Forest', 0.8612596553773024),
  'Naive Bayes': 0.8324420677361855,
  'Random Forest': 0.8612596553773024,
  'SVM': 0.8523469994058229}}

Deployment

Add additional notes about how to deploy this on a live system

Built With

• Sci-kit - The framework used for the models
• Pandas - DataFrame framework
• Matplotlib - Library used for plots

Authors

• Ana Estrada - (https://github.com/AnaCReal)
• Emilio Ferreira - https://github.com/efferreiram)
• Rodrigo Careaga - (https://github.com/lrodrigocareaga)

See also the list of contributors who participated in this project.

License

This project is licensed under the MIT License - see the LICENSE.md file for details