"A Project is complete when it starts working for you rather than you working for it" - Scott Allen
Ecommerce Project
An Online business was having issues in increasing sales
After doing the root-cause analysis, it was observed that the business lacked a good market-mix-model
Developed models and provided with suggestions for increasing sales
Time Series forecasting
There was a need to forecast sales for a retail giant and also suggest the best method of forecasting.
Applied various methods of time series forecasting and provided with suggestions for best method which reproduced the trend and seasonality components with least error
Bike Sharing
A bike-sharing provider recently suffered considerable dips in the revenues due to the ongoing Corona pandemic hence decided to understand factors affecting demand for bikes
Designed Linear Regression model to observe the highest linearity and factors affecting the rentals.
Identified top factors with positive and negative correlations that will help in business
Leads Scoring Case Study
For an online education company, it was observed that the conversion rate for leads was quite less.
Hence built various models and calculated
probability statistics to give scores to the leads, so as to tap the hot leads for conversion
Heading 6
Designed Logistic Regression models with an accuracy of 0.81 on train-set and 0.76 on test set and AUC of 0.87. Key Achievement: Segregated leads based on probabilities and gave suggestions as to which leads should be contacted first so that the prospect leads are not lost
Deployed K-means and Hierarchical Clustering to categorize the countries based on socio-economic conditions and reported the list of countries in most dire need of financial aid.
Key Achievement: Achieved a Hopkinns score of 0.91 and 3 distinct clusters in complete linkage type. Presented the results with visual plots.
Optimization
The decision variables, objective function and Constraints were defined using pyomo as per requirement and various combinations of staff and outsourcing percentages were checked for achieving staff optimization.
Consumer complaint resolution
The consumers who are likely to dispute were predicted so that they can be given more attention as to how the complaints are handled.
Exploratory Data Analysis’ on dataset ‘Indian Premier League’
To find out the most successful teams, players and factors contributing win or loss of a team.
To Suggest teams or players a company should endorse for its products.
