Demand forecasting
Demand forecasting is an important business task and if you want to reduce the time spent and increase the accuracy MLmargin is the right tool for you! Artificial intelligence for supply chain management.
What data does it use
MLmargin requires historical sales information for all the items you want to get a prediction for. Minimum one year of data history is preferred for an optimal accuracy. There are two options for data input:
- Upload it to form your computer
- Let MLmargin fetch the data by using API integration with your already available management software
This sales data will be used to train the machine learning model.
Important are the following attributes from the sales history: Item ID, date, quantity sold, price.
What do you get
Forecast periods: select the time frame that works best for your planning needs
- daily
- weekly
- biweekly
- monthly
Lean design: you can drill down on item level on our UI for a detailed view. This supports your analysis and decision taking
Seasonality recognition: if you sell items that are seasonally affected such as fashion this has to be taken into consideration. For that reason we use machine learning feature extraction or deep learning. This way the AI can reveal the hidden potential of your data and provide accurate results.
Trend recognition: whether your company and revenue are growing or decreasing linear interpolation is used to bring trends into the final prediction result
What technology does it use
We know that every customer has different supply chain management data characteristics. They depend on the business growth stage and the technology for recording and storing the data.
Thus it is important to have a variety of techniques which perform best on all possible combinations of data such as big data or small datasets and sparse or dense data. MLmargin has technologies adaptable to any data scenario.
It leverages the flexibility and the big data capacity of Deep Learning and the robustness and stability of standard Machine Learning tools. We have combined these modeling techniques within the concepts of Reinforcement and Transfer Learning in order to make them solve the growing optimization challenges of the SMEs in the retail and ecommerce industry. Read more…
Replenishment optimization
It is challenging to maintain optimal inventory levels in always changing business conditions, customer preferences and product assortments. Tools based on Artificial Intelligence provide more accuracy and profit increase. MLmargin: the all-you-need AI tool for inventory optimization
What data does it use
This optimization model relies on the results of the demand forecasting machine learning model, lead times, current inventory level, already ordered quantity, price
What do you get
Recommended order quantity for each SKU: The machine learning model generates and examines a huge variety of possible scenarios and gives you the best recommendation to achieve your business goals. It considers both lost revenue caused by stock outs and the cost of unnecessary high inventory levels.
By replenishing the correct quantity you can reduce delivery time to end users thus improving brand reputation and customer loyalty.
Supply chain management decision support tool MLmargin is the right choice for proactive and innovative retailers
What technology does it use
MLmargin uses probabilistic optimization algorithms powered by reinforcement learning. Reinforcement learning is a very popular technique for Artificial intelligence.
It trains machine learning models to make a high number of decisions in order to achieve the best result in an uncertain environment. It is composed of 3 main features: agent, action and reward. The Artificial intelligence learns to achieve a goal in an uncertain, potentially complex environment. The agent gets either rewards or penalties for the actions it performs thus leading the model to the best possible result.
