Data Science Portfolio

Repository containing portfolio of data science projects completed by me for academic, self learning, and hobby purposes. Presented in the form of iPython Notebooks, and R markdown files (published at RPubs).

Contents

• Machine Learning

  • Personal Loan Acceptance - The aim of this project is to help Universal Bank to convert there liability customers into loan customers. In this study four classification algorithms (Logistic Regression, Naive Bayes, Support Vector Machine and Random Forest) were trained. Out of all trained classifier Random Forest outperformed with PR AUC Score of 0.955 and F1-score of 0.951.

  • Portugal Bank Marketing Campaign - In this project, we will evaluate the performance and predictive power of a model that has been trained and tested on data collected from customers of portugal bank. Varoius classiffication model we trained on this dataset but KNN outperformed other algoirthms with AUC Score of 0.829841 and F1-score of 0.829841.

  • Credit Card Customers Segmentation - In this project, we need to extract segments of customers depending on their behaviour patterns provided in the dataset, to focus marketing strategy of the company on a particular segment. The aim of this analysis is to group credit card holders in appropriate groups to better understand their needs and behaviors and to serve them better with appropriate marketing offers.

  • Mobile Price Prediction - In this project, we have considered a multiclass classification problem where we need to predict price range of a mobile. The dataset given is balanced with respect different categories of target feature (price_range). We built a Gradient Boosting Classifier which outperformed other classifiers with test accuracy of 91.8% and a f1-score of 0.917.

• Time Series Analysis

  • Global Climate Change Analysis - In this project, we have performed in-depth analysis to study the change of climate across all many years. Also we have built a Seasonal ARIMA model capable enough to forecast temperature of bombay city. According to the forecasting, Bombay will record a highest temperature of 28.55ºC in the month of April i.e during summers. Monsoon is going to be cooler and there will increase in temperature in post-monsoon period. The temperature in winter's will remain same i.e 25ºC.
  • Sales Forecasting - In this project, we have performed in-depth analysis of the sales of a store with respect to segment of customers, region, city, product's category, sub-category and the month of the year. The store generated the total-revenue of $209,624 in the time stamp of four years. The time-series model was built to forecast the sales of the store after seven days. According to the forecasting sales of the store will be $ 3060 on 2018-01-06.