Yash Agrawal

Binary Classification Problem to predict and forecast incident happening or not at CTA locations using Various Significant Attributes.

Contributions:

• Fetched and developed the train data using Here API, implemented Feature Selection using Filter method - Chi2 using Python
• Used correlation matrix to evaluate the correlation among variables and then used N-fold cross evaluation.
• Developed Predictive models from classsification algorithms Adaboost, Bagging and gradient Descent and random Forest.
• Evaluated and Measured Metrics with test data in Accuracy, Precision and recall and got Accuracy of 79% in test data.

Implemented Fault Prediction Model to Forecast happening of network fault and what type of fault was that.

Contributions:

• Used R script to import data from telecom network, performed cleansing and preprocessing data using Power queries.
• Used correlation matrix to evaluate the correlation among variables and then used N-fold cross evaluation.
• Interpreted the Mining frequent pattern using sequential pattern discovery using equivalence classes (SPADE) using R studio.
• Evaluated and Measured Metrics with test data in Accuracy, Precision and recall and got Accuracy of 79% in test data.

Data Warehousing (DW) is a process for collecting and managing data from varied sources to provide meaningful business insights. Data Warehouse plays an important role. It helps to analyze key aspects to improve sales of an enterprise.

Contributions:

• Designed an end-to-end sales data warehouse using Pentaho DI and performed various transformations on the data (Row Normalizer, Database Lookup, if field value is null)
• Designed a database using Star Schema and built a multi-dimensional cube using Mondrian
• Configured the Pentaho BI server to deploy reports by creating a database connection in Pentaho Enterprise Console for central usage.

This is Bike sharing prediction for forecasting the Sharing business of bike over the USA.

Contributions:

• Applied Data processing of data and feature Engineering forgetting insights, Implemented Residual Analysis on developed models
• Developed Linear Models on different parameters metrics AIC/BIC , Applied ANOVA on the relational Databases.
• Trained model with methods such as Random Forest , Support Vector Machine(SVM) algorithm got accuracy of 87% on test data.

Have you ever used a credit card at a store only to be declined? Or have the payments been blocked because you were charged a higher amount. Credit Card Fraud is a wide-ranging term for theft and fraud committed using a credit Card or any similar payment mechanism as a fraudulent source of funds in a transaction

Contributions:

• Built a model that identifies the factors influencing the world happiness with 92% accuracy
• Used classification algorithms like KNN, Naïve Bayes and logistic regression to validate the results.
• Performed data cleansing and preprocessing using R statistical packages.
• Identified significance variable by Weight of Evidence(WOE) and IV analysis, to solve imbalance issue where 0.17% are fraudulent.
• Evaluated developed data model by machine learning algorithms a Logistic Regression, Decision Trees and Random Forest to predict Fraudulent Transactions and validated the models using confusion matrix and F1 score.