/Virtual-Internship-Experience

Data Science & Analytic Virtual Internship Program from Boston Consulting Group, Accenture, KPMG, and many more

Primary LanguageJupyter Notebook

Virtual-Internship-Experience

Virtual Internship project i got from Boston Consulting Group, Accenture, standard bank, KMPG, and Cognizant

Boston Consulting Group : Churn Probability Model and Business Simulation

This goals of this project are predict probabilty of customer churn in clients company and find the best cut off treshold to maximize potential revenue

Data Insights:

  1. All price features have low correlation to target (churn)
  2. Our client have approximately 10% churn customer
  3. All consumption features are positively skewed
  4. There are no high correlation among independent features to target feature
  5. The highest correlations on target feature are just 0.10 on margin feature

Recommendation:

  1. Due to no high correlation to target feature, if possible we need additional customer data (customer profile, interaction, and behaviour)
  2. Need external general competitive price on market pricing, because we want to know if its make sense/effective to give discount 20% to churn customers

Model Insights:

  1. Discount Strategy Discount strategy of 20% is effective but only if targeted appropriately. Our recommendation is offer discount to only to high-value customers with high churn probability

  2. Feature Importance: Top drivers of client churn are antiquity of the client, tenure, and difference off peak - peak in energy price. We can say price sensitivity is main driver for clients churn, but not the most important factor

  3. Potential Revenue: Maximum benefit at cutoff 0.22 with revenue delta of $60,949.26

Accenture : Audit Sample Big Data and Recommendation for succesful IPO

Client Request:

To start our engagement with Social Buzz, we are running a 3 month initial project in order to prove to them that we are the best firm to work with. They are expecting the following:

  1. An audit of their big data practice
  2. Recommendations for a successful IPO
  3. An analysis of their content categories that highlights the top 5 categories with the largest aggregate popularity

Insights Summary:

  1. Computer Generated Data There are some columns suspected as computer generated data. Because some of columns are constant and don't represent a pattern
  2. Most Watched Content Type We just can provide insights about most watched content type, highest positive and negative sentiment score, etc

Recommendation

We need need additional data, during analysis we realized that there's something wrong on the data that client's provided. Client need to evaluate the data team about how they generate the data. Our recommendation for this case is data team and data environment evaluation.