Shubhro Sankar Basu

Data analyst having a background in computer science and an eye for working with numbers and data. Here to develop solutions and solve problems using analytics and enhance existing systems. Have significant working experience in reconciliation, workflow modelling , cloud platforms and analytical tools. Enthusiastic about using technology and tools to confine, deal and handle large volumes of data effectively and efficiently in today's highly data-centric world

1)Rainfall Trend Analysis and Prediction in India's Western Ghats Using Sen's Slope and Machine
Learning Algorithms

• Addressed significant research gaps in the study of climatology in the Western Ghats, a biodiversity
  hotspot in India.
• Employed Sen's Slope estimator to analyse historical rates of change in monthly rainfall, offering rare
  temporal comparisons for evolving climatic patterns.
• Used Sen's Slope technique to identify anomalies and trend reversals in monthly precipitation,
  adding a new layer of granularity to regional climate studies.
• Conducted a novel comparative analysis between traditional machine learning algorithms (Linear
  Regression, Random Forest, Gradient Boosting) and time-series models like ARIMA to forecast annual
  rainfall.
• Aimed to fill existing research voids and provide ground-breaking insights into climate dynamics
  crucial to India's ecology and agriculture.

2)Advanced Criminal Behaviour Analysis Through Statistical Modelling in R

• Conducted a comprehensive research project using R programming to analyse a complex dataset
  on criminal behaviour.
• Undertook meticulous data cleaning to remove missing and inconsistent observations, setting the
  foundation for robust analysis.
• Engaged in exploratory data analysis, creating numerical and graphical summaries that segmented
  the data by various motives for criminal activity.
• Proposed and evaluated different statistical distributions to model the age at the first kill across
  multiple motive categories.
• Enhanced skills in data cleaning, statistical modelling, and inferential analysis through hands-on
  experience.

3)Predictive Modelling of Boston House Prices using Extreme Gradient Boosting Linear Algorithm

• Led an initiative focused on advancing predictive analytics in Boston's housing market by building
  an Extreme Gradient Boosting Linear (xgbLinear) model, utilizing the well-known Boston House
  dataset.
• Executed meticulous data pre-processing activities, including variable normalization and strategic
  splitting of the data into training and validation sets, aimed at enhancing model performance and
  validity.
• Applied hyper-parameter tuning techniques to refine the xgbLinear model, achieving robust fit
  measures that confirm the model's accuracy and general applicability.
• Assessed feature importance through the finalized model, gaining insights into key determinants of
  housing prices.
• Justified analytical choices such as data normalization and validation splits, specifically geared
  toward optimizing the model's predictive power and reliability.
• Contributed to the field of real estate analytics by demonstrating the efficacy of gradient boosting
  algorithms in accurately predicting housing prices.