Data Science in Supply Chain (Optimizing Forecasting and Planning parameters)

Scope

Data science in Supply Chain is a very big vast area.  For this article, I would like to focus on Optimizing Forecasting and Planning parameters in Supply Chain using Data Science.

Contents

Data science in supply chain is being used for a long time and software products are available in the market in different names.  However, my focus in this article is about the business opportunities in the Supply Chain and Data science area for better configuration of parameters.

Every company while implementing Supply chain software products for forecasting and planning, they have to configure bunch of parameters. These parameters play a crucial role in the optimization of forecasting and planning algorithms.

Now, what are the optimum values for these parameters that would result in best supply chain optimized output.  Shall we write another optimization algorithm to optimize the configuration parameters for the optimization of forecasting and planning algorithms? Not really. Then, how do we set optimum values. This is one of the area where Data Science could be utilized.  Before getting into the details, let me give a very high level overview of the Data Science relevant to this article.

Data Science overview

Data Science has 3 components, viz. Hack skills or Programming, Mathematics or Statistics and Substantive or Domain Expertise.  It is possible one Data scientist may not have all of these skills.  It would be a combination of one or more people to work out all these components for any project work.

Also, let me give an overview of the typical Data Science project (once again, I scope this relevant to this article).  The goal of the project is set and align with the team consisting of one or more Data scientists.  Set of data needed for the work is determined.  Data input, validation and cleaning done.  Data exploration and model creation done using statistical methods.  Model applied, validated.  Output of the model shall give the recommendation to satisfy the goal of the project.  Of course, all these are documented and archived properly.

Project work or business opportunity

For our context, to find out the optimum configuration values for the forecasting and planning parameters, first determine the top 10 parameters that would contribute to an optimized forecast and plan.  

• Determine the goal for each parameter
• Create a Data Science team - 1 or 2 Data Scientist with skill set covering 3 components of Data Science, Business Expert, Project manager
• For each parameter, determine data input needed.  Most likely, these would be associated with Forecast, sales, etc. or inventory, plans, etc.
• Data Scientists shall analyze the data, do programing, create statistics models, validate and determine the optimum parameter value.
• The final output shall be reviewed by all stakeholders to ensure desired goals are achieved
• Apply the model in production
• Validate the results in periodic interval; if further tuning is needed,
• follow the above cycle
• Repeat for each parameter

Data Science Project in Supply Chain implementation

Each parameter optimization could take about 4-6 weeks initially as this is a new area for Data Scientists.  Once they become expert in analyzing this, it could be done in 2-3 weeks.

In about 6 months, a team could make a better return for the company.  It can be repeated in future any time, if needed.

Summary

Nowadays, lots of data scientists are available.  Universities are teaching and enabling students to do project work like this.  As needed, many such short term projects could be done to improve the supply chain configuration parameters.