Inventory Management

Brief

Saudi basedOEM had manufacturing units in multiple location in Saudi Arabia. However, it was witnessing lot of issues with inventory planning including higher inventory cost (compared to its competitors) and regular stock outs.

Challenge

During our initial analysis, we discovered that safety stock levels & order replenishment levels were not backed by sufficient data. So there were times when safety stocks levels were not sufficient to prevent stock out and on the other hand there was set of inventory which kept lying in the ware house for months without consumption. Moreover, there was no central system to keep a tap on inventory levels and consumptions.

Our Action

• We collated the historical sales data and  gathered the external factors  for the period under study including seasonality and macro indicators
• Performed the data warehousing and data mining to get the data cleaned and ready for the model
• We  trained the machine  learning model (with XGBoost algorithm)to get to the category wise demand forecasting
• As per the model results, we redefined safety stock levels & order replenishment levels 
• We additionally  created the Comprehensive dashboard for management to have visibility  on real timedemand and inventory for different categories

Results

• Stock out situation reduced due to improved demand forecasting
• Accurate safety stock levels reduced inventory cost without risking stock-outs