Global Data Science Forum

Expand all | Collapse all

feasibility of a project

  • 1.  feasibility of a project

    Posted Sun December 01, 2019 06:49 PM

    Hi there 🙂🙂..Can anyone help me?

    I want to create, for my bachelor's internship, a software or a Web application making it possible to increase the profitability of a product (ex: sale of software) ....

    the theme is "profitability intelligence of a product"

    Here's what I found on the net to achieve this:

    1-Customer Intelligence (profiling)

    2-Targeted Marketing

    3-Competitor Intelligence

    4-decision for promotional prices

    Question 1: Is the theme not very broad to cover the 4 points mentioned above for a beginner level ?? ... or should I create a software / application that only covers one or some of these 4 points?

    Question 2: Can you advise me to prepare my dataset and machine learning algorithms needed for the project?..🙂🙂🙂...

    thanks for answering

    Luis Zefania

  • 2.  RE: feasibility of a project

    Posted Tue December 03, 2019 01:37 AM
    1. This looks fairly broad as it covers customer analytics and marketing analytics theme. If the theme is "Profitability intelligence", some of the use case theme could include - Customer Life Time Value (CLTV) analysis, Customer Churn prediction, Clustering of products and targeting NBOs (Next Best Offers). Loyalty Analytics, Campaign Analytics etc. along with earlier mentioned themes are also related. I think it would be better to focus on a particular theme around customer analytics and ask business goals that it can solve and how to accomplish them leveraging Data Science and how they help in improving profitability for the customer. (can be across any industry for that matter based on data relevance and problem at hand).
    The goal of profiling customers that are vulnerable to churn and devising retention strategy to those customers for retention could be a good focus point to deal with. The problems / strategies that customer would look for in this case could be as follows: (just illustrative) 
    a) Ability to offer personalised product and services to customers to improve customer satisfaction and brand value
    b) Identify loyal customers and target specific promotions to retain them
    c) Reduce customer churn and thus manage risk

    I think it would be great to put together a holistic framework and explain the same. That means the journey or roadmap would be from "Customer Acquisition" to "Customer Development" to "Customer Retention".

    Determining prices or Pricing analytics etc. are again huge area within themselves.

    It is always better to put together a business goal/theme, put together 2-3 key business questions that solution will solve and then accomplishing that with multiple approaches, experiments etc to showcase what has been done to achieve the same. (and why the chosen or recommended experiment is picked as best solution) and so on..

    2. From datasets standpoint, it would be good to leverage from some open datasets that can be used for similar problems. Key would be to define KPIs, to understand data by performing EDA and showcasing visualisations, formulate different attributes/features or to prepare for feature engineering, then multiple experiments to showcase different algorithms tried, and to choose the algorithm or experiment based on specific model evaluation approaches.

    a) Google datasets
    b) Kaggle datasets
    c) US govt open datasets for usage
    d) UCI Machine Learning Repository
    e) Quandl

    Above are some examples / references. Please use your discretion / check to see which one can be used for learning purposes or equivalent. If the objective is not on big data and more on machine learning, then it is better to look at some sample size and leverage something which has more features, more variations in it's historical data etc.


    Kamal Mishra