Dr. Anuj Bajpai

Highly organized and motivated Data Scientist with expertise in gathering, cleaning, and organizing data for both technical and non-technical stakeholders. Advanced proficiency in statistical, algebraic, and analytical techniques, coupled with a significant background in machine learning algorithms, including theoretical and applied aspects. Skilled in computer vision and deep learning methodologies, adept at leveraging these technologies to extract insights and drive decision-making processes.

Project 1: Sales Forecasting Using time series data

Problem Statement : To suggest possible forecasting models for prediction of shipped units for a list of 110 products

• Developed a precise and well-defined comprehension of the business requirement.
• Figured out anomaly in the data using proper EDA techniques at the early stages of the project.
• As data was having a superimposing random component and all statistical models were not working satisfactorily. In such a condition prepared an exhaustive list of experiments where most of the possible models have been tried for wide range of products.
• Applied a variety of models, spanning from traditional to cutting-edge, in the realm of sequence modeling.
• Establishing a firm performance criteria for different models such as to decide optimal performance of a model.
• Achieved better forecasting accuracy compared to baseline models or industry-standard methods, showcasing the effectiveness of the chosen approach.
• Tuned model hyperparameters systematically to enhance performance, demonstrating a strong grasp of model configuration.
• Delivering results at stipulated time with well prepared documentation.

Projects 2: Hr Suite Attrition prediction.

Problem Statement: The main aim of work was to develop an accelerator for prediction of attrition(turnover), termination reason, turnover reason and type as well as forecasting attrition using time series forecasting.

• Learned as well as implemented various machine learning techniques like feature selection, k-fold cross validation techniques, SMOTE, ADASYN(data augmentation techniques).
• Tried, tested and tuned hyper-parameters for different machine learning models like logistic regression, decision tree, SVM and marked the best possible.
• Successfully accomplished 4 tasks out of 5 with commendable accuracy, and other parameters like precision, accuracy and F1 score. Also made use of kappa coefficient in case of ties between models.

Project 3: Optimization Singular Control Technologies

Problem Statement : A well structured python program making use of efficient libraries for solving operation research problems is desired. Developed application should be equipped with capability of solving linear/nonlinear optimization problems as well as iterating over multiple solvers.

• Expertly formulated the problem as a mathematical optimization model, translating real-world constraints and objectives into a format compatible with OR-Tools.
• Designed custom objective functions that appropriately quantified the problem's goals, balancing multiple criteria to achieve the desired outcomes.
• Made use of pulp library and goggle OR tools for solving for solving constrained optimization LPP ans QPP.

Project 4: Cohort analysis for ride business.

Problem Statement: To perform cohort study, churn analysis/retention analysis/personalized customer experience.

• To demonstrate our expertise for the bidding project, we will commence by generating the necessary data and then proceed with the full implementation of the analysis outlined above.
• Created an extensive dataset containing trip records, customer profiles, and driver information, ensuring seamless interconnection and consistency among all the data components.
• Designed an impressive dashboard spanning five pages that effectively presents analytics derived from trip, customer, and driver data.

Python for Data Science, AI & Development Course IBM Python for Data Science, AI & Development