Project information

  • Category: Classification
  • Project date: March, 2019
  • Project URL: Github
Photo by Christian Fregnan on Unsplash

Finding Donor for Charity using Machine Learning

Finding donor for Charity ML project is trying to find an algorithm that best identifies potential donors. This project is completed under Udacity Data Scientist Nanodegree Requirement.

Project Requirements

Perform necessary steps to acheive a higher donation yield rate.

  • Preprocess the data.
  • Create visualization like data distribution.
  • Perform transformations on features that are highly skewed.
  • Evaluate and optimize several different supervised learners to determine which algorithm will provide the highest donation yield.
  • Describing the Model in Layman's Terms.

Skills Required

Python
Pandas
Matplotlib
numPy
LogisticRegression
Support Vector Machines (SVM)
GradientBoosting Classifier
AdaBoost Classifier
GridSearchCV

Techniques

Major Tasks

  • Creating visualization to explore features distribution.
  • Normalizing highly skewed features using MinMax scaler.
  • Using Onehot encoding to preprocess categorical variables.
  • Comparing results of Logistic regression, Support Vector Machine(SVM) and Gradient Boosting Classifier to predict possible donors.
  • Adaboost Classifier is also used as part of the ranking most important features from training data. The final Model has been tuned using a grid search(GridSearchCV) functionality.
  • Explored feature selection to achieve similar level of prediction accuracy with only 5 top features.

Completed using Jupyter Notebook.