|dc.description.abstract||Effective businesses implement a monthly Sales and Operations Planning (S&OP) process within their organizations to align and coordinate different functions of the business under a single plan. Despite this significant undertaking, once businesses align on a S&OP plan, little effort is made on methodically assessing risks in the plan to determine opportunities for improvement.
The objective of this project was to help a sponsoring company in the Food & Beverage industry proactively predict high probability risks in their S&OP plan (i.e., imbalance of supply and demand), allowing them to mitigate the negative consequences that result from reactively managing these risks. Data mining methodologies were applied to S&OP data for a selected company brand with the goal of determining model(s) that could best predict high probability risks for intervention and risk mitigation. The predictor (independent variables) and response (dependent) variables were derived from the S&OP data and the models were trained, validated and tested in R. Supervised classification algorithms were used to build classification models for each of the four risk outcomes (binary); 50% over forecast, 50% under forecast, weeks of supply below four weeks, and stockouts.
Of the four risk outcomes studied, only 50% over forecast provided a viable model for business application. Compared to business as usual, this classification model (applied to a 12-week period) improved 3-month lag accuracy by 5.7%, reduced bias to near zero and added a conservative $415,000 in operating profit. Applying methodologies identified in this study across all brands and extrapolating operating profit improvement across a full year, the sponsoring company can capture $15MM increase in profits annually. This study concludes by recognizing that even without large data sets (i.e., big data), there are a multitude of benefits companies can gain through the application of predictive analytics to capture business risks in their S&OP plans.||en_US