Throughout the past decade, Pluto7 has spearheaded many predictive analytics projects for multiple fortune 500 companies. We have found that two common question continue to come up in various Supply Chain Planning projects. The first question is "Can you explain the peaks and valleys in the forecast curve?" In a few isolated cases this may be relatively easy to explain, but in most cases the answer is far from straightforward. The second most common question is "How can we improve forecast accuracy?" Both of these questions are very reasonable and expected from lients to ask when they invest significant amounts of money on these projects and hence expect a very reliable output.
However, answers to these questions really rely on three things:
A. Good quality data
B. Appropriate business rules
C. Reliable foundational infrastructure
Getting a handle on A and B is much harder than it initially appears. In this journey of trying to uncover the best answers, we ended up designing a SaaS solution called Planning in a Box. We have decided to address this common supply chain planning conundrum by building on a cloud foundation (e.g. solving for C) what is commonly known as MRP - or a planning engine for supply demand balancing..
It is our observation Analytics industry is currently going through at an accelerated pace of consolidation from vendors to solutions. Customers now expect a more integrated subscription-based solution and, interestingly, customers want to start and finish their entire analysis and associated transactional activity all within a browser window.
Thus, a good integrated future-ready Supply Chain analytics solution must have:
Step 1: Quality, intuitive visualization
Step 2: An easy to use data preparation tool
Step 3: Secure Cloud (or private Cloud) based Data Warehouse that scales
Step 4: Machine Learning features that are already IoT friendly
Step 5: Minimal Management requirements
With these being the customer expectations, we chose Paxata for data preparation ( solving for A above). With any supply chain solution, the quality of the input data is critical E.g. Forecasts, Sales Orders, Supply Commits and more. Having the right end user friendly tool on a browser is critical and Paxata provided us the right platform. Additionally, having the right visuals supported by the industry leader Tableau, embedded within the Planning in a Box application, users are now able to clean the data, load the data and create the dashboard, all within a single browser session without anything installed on their machine. The most amazing part is that the backend scales infinitely without any performance impact for any data volume or user loads by leveraging Google Cloud Platform.The relevance of this kind of fully-integrated solution increases when you start factoring in Machine Learning
Algorithms (commonly known as Artificial Intelligence) and IoT data into your Supply chain planning models.
Machine Learning is a paradigm shift for analytics. Business users no longer need to stop the flow of work to define business rules and rather let the machine identify data patterns and thus recommend the key data to focus on (e.g. Top 10 customers, Top 10 shortage parts, etc., similar to the Amazon or Netflix experience). The most challenging aspect of business adoption of machine learning is defining the right model. However, once initially defined, your ML model will continously learn and calculate far better than human limitations allow for. When it comes to forecast accuracy, it is all about constantly looking for obvious and not so obvious patterns and then adjusting your forecasting model on the fly. Humans at some point get exhausted scanning for patterns, fixing flaws and refining the model, hence the dreaded forecast inaccuracies in the supply chain continues to persist throughout the entire value chain. With Machine Learning, the machine will never get exhausted and can identify patterns in your demand and supply fluctuations which are practically impossible for humans who are generally too busy with supply demand balancing daily activities e.g. safety stock settings and more.
A Machine Learning model takes away the burden of defining Business Rules (Solving for B above).
CONCLUSION: The aspect that we cannot ignore as we evolve with these model is the cleanliness of the data and defininging the right model (predictive or machine learning). At Pluto7, we are constantly studying the market and defining the newest use cases that can transform the supply chain industry, working with our key cloud partners like Google and Fortune 500 customers.
We encourage you to stay in touch by subscribing to our blog. If you are interested in experimenting for free with our planning in a box solution, please submit your request for a trial license here.