Lokad optimizes supply chain decisions to maximize financial return using Quantitative Supply Chain principles. This financially-driven perspective focuses on reducing dollars (or euros) of error. Lokad’s recommendations, such as purchase or allocation lists, are balanced with respect to your specific supply chain needs and constraints - all of which is factored by our Supply Chain Scientists using probabilistic forecasting and Lokad’s domain-specific language (DSL) named Envision.
This video tutorial demonstrates Lokad’s public demo account, whiteboxing the optimization process and showcasing the predictive power and flexibility of our solution.
00:00:00 Intro to Lokad and Quantitative Supply Chain
00:02:16 Chapter 1: How we prepare the data.
00:03:23 Low-level data health checks
00:04:36 High-level data health checks
00:05:25 Chapter 2: How we use probabilistic forecasting.
00:08:05 Explanation of demand-over-lead-time probability distribution
00:09:18 Differentiable programming and its role in demand forecasting
00:10:02 Chapter 3: How we optimize supply chain.
00:10:30 Explanation of key economic drivers
00:11:25 The importance of the basket perspective in stocking strategies
00:12:12 The logic behind Lokad’s purchase recommendations
00:18:01 Auditing tools available for the client
00:19:50 Additional learning resources
The video is divided into four parts, designed to clearly guide the viewer through the various steps of a supply chain optimization project with Lokad. Below is a short summary of each part with additional links for further learning.
Data Collection and Preliminary Checks
Data pipeline: Lokad meticulously designs an automated data extraction pipeline, which facilitates the seamless transfer of vital data from the client. This data encompasses essential metrics like sales history, stock levels, and supply chain constraints.
File transfers and types: To ensure versatility in data transfers, Lokad’s platform supports a range of file transfer protocols, notably SFTP and FTPS. Beyond the transfer methods, the platform is adept at processing a variety of file formats including plain text, CSV, TSV, and Excel.
Data health: Lokad takes great pride in guaranteeing the health of client data, and invests significant time ensuring it is consistent and error-free. This is accomplished through the performance of a two-tier data health-check process.
- Low level data checks: This ensures data integrity and checks for discrepancies both within tables and between tables.
- High level data checks: This confirms data reflects the client’s primary KPIs as well as any business-specific features that will likely impact demand forecasting and overall optimization.
Security considerations: From a security standpoint, Lokad prioritizes client trust; clients maintain control over the access rights to their Lokad account. For those interested in a deeper dive into data protection, Lokad offers a comprehensive Security FAQ detailing its confidentiality protocols.
Forecasting Future Demand
Probabilistic forecasting: Traditional methods (such as time series forecasts) focus on predicting a single value, but Lokad uses probabilistic forecasting, which projects all possible sales trajectories and their likelihoods. Probabilistic forecasting is how Lokad contends with the irreducible uncertainty of future demand - something that a traditional time series cannot.
Differentiable programming: Supply chain management is marked by constantly shifting patterns, and to navigate this Lokad leverages advances in machine learning (ML) to continuously refine the numerical recipe it uses to generate decision-recommendations. Fundamental to this is differentiable programming - a method where the model continuously refines the forecasting technique with the emergence of fresh data. This allows the numerical recipe to evolve by learning from past data.
Translating Forecasts to Supply Chain Decisions
Economic drivers: Lokad’s quantitative supply chain approach is capable of handling vast arrays of economic drivers, even counterintuitive ones like stockout penalty (which is actually a reward driver). These drivers help Lokad to quantify the financial implications of various supply chain decisions.
Basket perspective: Lokad believes that stockout events for some SKUs carry unexpectedly high financial impacts, and these are disproportionately high relative to their direct margin contributions. In other words, some items, such as fridges, are typically bought in isolation. Others, like milk and bread, are typically bought in baskets, i.e., in combination with other goods. Thus, the unavailability of certain SKUs may influence a customer’s overall purchasing decisions.
Ranked decision-making: Lokad balances the probabilistic forecasts and the unique supply chain constraints of each client in order to generate prioritized lists of supply chain decisions. These lists are ranked in terms of the ROI generated for each one at the relevant level of granularity.
Supply Chain Performance Monitoring
Dashboard monitoring: Lokad provides extensive interactive dashboards to help end-users unpack and comprehend the lists of decision-recommendations. These tools help the client visualize the probabilistic demand forecast, stock trajectories, financial prioritizations, and how their supply chain constraints are factored into the optimization process.
Proactive analytics: The client has constant access to detailed, real time analytics - including tracking stock trends, out-of-stock percentages, lost sales estimations, and potential divestment opportunities - so they can constantly maximize their return on investment.
Alternatively, you can explore our testing platform at try.lokad.com. There you can utilize your own data and code your first supply chain optimization scripts.
For long-form research articles on supply chain theory and best practices, as well as downloadable study resources, be sure to review our extensive Learn section.
For in-depth lectures on vital supply chain concepts, interviews with expert guests, and short explanations of industry terms, visit LokadTV.