- Fraud Detection Data Analysis -
Fraud Detection dataset project with a aim to implement various data analysis and machine learning modeling techniques. Following are the components in notebook:-
- Data Glimpse - Data Distribution of some of the continuous variables, Target variable distribution along with other variables, other EDA.
- Data Preparation - Balancing the unbalanced dataset, Data chunking.
- Data Preprocessing - Missing Values Treatment, Outlier detection & Capping, Feature Engineering & Data Encoding.
- Data Modeling - Baseline Random Forest, Adaboost (Ensemble), Linear SVC. Accuracy measures - AUC_ROC curve, Classification Matrix.
Model Inferences and Conclusions in the end notes of py notebook.
- Gradient Descent From Scratch -
Implementation of Gradient Descent using Logistic Regression from Scratch for Predictions. Gradient Descent is a first order iterative optimization algorithm fro finding the minimum of a function. Gradient descent is also known as steepest descent. Data could be manipulated in the "main.py" file and the same file could be run for Predictions.