Wellcome
Welcome to my data science portfolio! In the folders above are various data science projects that showcase my skills in wrangling/munging, manipulation, visualization, analysis known as EDA, experimenting and modeling using Machine Learning techniques with data.
Following is a brief overview of the various projects those I completed. More details can be find in each respective folder.
Capstone Project New York City 311 Non Emergency Complaint System
Project Overview: The people of New Yorker use the 311 system to report complaints about the non-emergency problems to local authorities. Various agencies in New York are assigned these problems. The Department of Housing Preservation and Development of New York City is the agency that processes 311 complaints that are related to housing and buildings. In this project I developed a predictive model for a future prediction of the possibility of complaints of the type that have identified
Skills Showcased: EDA, Inquisitivity, Statistics, Visualizations, Research Outcomes and Recommendations
Project Overview: The commerciality of unconventional field development heavily lies on the ability to generate highly productive stimulated reservoir volumes (SRV) in an economical manner. However, there is no reliable method to estimate SRV consistently in the course of or immediately after fractur-ing operations. The amount of SRV achieved during fracturing can be inferred through multiple techniques, including production data analysis, modeling (analytical or numerical, including var-ying treatments of geomechanics) and/or microseismic monitoring. All these approaches have several limitations.
Using production data, we need at least several months of well production history to establish pseudo-boundary dominated flow regime. Additionally, the production-derived total SRV is a composite of all the stages lumped together, as it is difficult to evaluate the performance of indi-vidual fracture stages due to the reservoir heterogeneity; furthermore, production rates are meas-ured as a total aggregate for all the fractures, while production contributions from each stage cannot be obtained pragmatically for most wells. Full physics modeling, besides being time con-suming and requiring specialized expertise, needs proper reservoir characterization including ge-omechanical parameters and calibration, which are not routinely captured for the majority of the wells, and even when they are acquired, they are difficult to accurately measure. Microseismic, though informative, doesn’t indicate actual SRV, and often it is only available for a small per-centage of the wells. It is desirable to have a scalable method to consistently estimate SRV evolu-tion during injection only using commonly available fracture treatment data for all wells, and practically applicable in a typical unconventional field.
The main objective of this work is to estimate stimulated reservoir volume during injection from routinely collected fracture treatment data in real time for each stage and compare aggregated stage-wise stimulated reservoir volumes to total well production drainage volume.
Skills Showcased: EDA, Data Cleaning, Feature Engineering, Data Driven Modeling using Machine Learning Techniques, Visualizations, Research Outcomes and Recommendations.
NOTE: As I developed this model for Anadarko Petroleum Corporation and the company owned the IP I'm unable to upload those code here. Just sharing those results as these are now in public domain