Delivery Time Forecasting for Supply Chain

Predict time to deliver time to reduce late shipment rate and uncover underlying issue early

End-to-end analytics and AI services

Pascal can help you to process large amounts of data from different systems in and across the supply chain system in an organization, transform and aggregate disparate data into one unified view, then apply machine learning with DataRobot to analyze the data, make prediction and draw out actionable insights.

Focus on performance against demand plans

Our ability to predict and forecast helps customers to take prescriptive actions and achieve their goals, where supply chain management has historically been reactive. By applying our experience and modeling tools, we can help customers proactively shape and direct their supply chain functions.

Enabling supply chain managers

Supply chain managers can become truly data-driven by leveraging data and machine learning model to better manage their supply chain functions to increase their ability to achieve their plans.

How it Works

The solution powered by Pascal Studio and DataRobot is designed to create a trusted, cohesive data pipeline and use machine learning to predict where supply chain problems might emerge. It provides businesses with an end-to-end view of supply chain performance by connecting data across your supply chain within your organization to provide early warning of potential disruption. 

In this solution, Pascal Studio helps customers consolidate historical supply chain, especially shipment data from numerous sources, verify data accuracy, adjusts for recent trends and create one unified view of all supply chain data in your organization. Pascal Studio also provides deep domain expertise in supply chains to identify exceptions that lead to impacted deliveries. For example, anomalies, off-trend data, lack of certain data from a particular supply chain function could all be signal for the machine learning model to predict potential issues.

DataRobot is used to analyze the data, predict delivery time for new shipments and individual containers, as well as where supply chain problems might occur. Using geolocation data, Pascal is able to train models that quickly learn about difficulties shipping to certain regions with DataRobot. 

Pascal also helps you to navigate common pitfalls during modeling process for such solution. For example:

  1. Creating the right target variable and training period to avoid selection bias due to shipment still in-transit, or building survival model using period-by-period binary classification.
  2. Properly using carrier’s estimated arrival time that are generated at the right time snapshot to avoid target leakage, and updating the estimated time of arrival accordingly before forecasting.

With this solution, customers can transition from constant firefighting mode to a predictive, proactive model of supply chain management by identifying potential problems early and providing the reasons causing these issues.

Key Deliverables

Improve data quality

We clean up the data that will be used for analysis, consolidating input from multiple sources.

  • Collect and evaluate the standard data signals that come in throughout the supply chain
  • Consolidate data from multiple locations, including third-party manufacturers, fulfillment centers, distribution channels, and legacy systems
  • Connect all fulfillment center and carriers’ last mile delivery data
  • Data normalization and anomaly detection to quickly identify data gaps
Create models to predict delivery time and improve on-time fulfillment

We combine our time-series capabilities and DataRobot to understand the relevancy of the data signals collected. We can forecast with DataRobot when materials will be delivered, and we can predict where issues may occur. 

  • Understand the flow rate across the end-to-end logistics network
  • Predict which shipments (down to the container) will cause customers to miss targets
  • Identify the biggest bottlenecks for shortages and disruptions
  • Data and model monitoring against plan at different time granular levels
  • Create executive dashboard

About the Partner

Pascal Studio merges supply chain domain expertise and data science expertise to drive and support successful digital transformation initiatives for customers. Pascal Studio helps you to connect data from internal and external systems across supply chain functions into one seamless view.

Here are services provided by Pascal Studio:

Data Quality & Machine Learning

  • Provide continuous data quality monitoring and reporting
  • Synchronize data across internal and external transaction systems
  • Anomaly detection and data improvement
  • Machine learning modeling and analysis

Inventory-in-Transit and Logistics analysis

  • Provide warning system to identify slowdowns and shortages in logistics
  • Identify excess cost from containers and inventory stalled through logistics routes
  • Identify routes with excess variability to reduce safety stock

Executive Dashboard

  • View live data in presentable format with customizable views to include multiple dashboards
  • Provide recommendations for action and adjustments to plans

Solution Diagram

Pascal solution diagram

Success Story

For one of the world’s top technology manufacturers, Pascal developed the use case to predict time to deliver for their shipment. The solution leverages information down to each shipping containers and geolocation data. The customer was also able to understand the drives behind each late shipment for different geo regions.

cta module 1920px

Ready to Get Started?

Take action now to transform your franchise organization with Ironside's data-driven decision-making solution.