Mathfi-SupplyChain© Classifier/Predictive AI

                    Log in/Register                 Request a Demo               Instruction Manual




A new AI-Powered helping your decision experts, or your AI-Powered Supply Chain platform make decisions by performing binary or multi-class predictions and classifications for example or sanity checking their existing predictions/classifications for example:


1. Predict the demand for a certain product to optimize the inventory: Butterfly AI can help businesses and manufacturers predict the demand for certain product during different seasons across any given region. Having the data features of thousands of past purchase events of the target products and past consumer details including consumers’ income, regional macroeconomics, regional inflation, customer trends and preferences, the level of marketing spent on that product, number of ads live on social media related to that product, number of TV adds live on TV related to that product, willingness to substitute the old product with the new version. durability of the product, product’s brand popularity, product's market penetration, product's online review scores, if it is close to the seasonal events such as Christmas, the region, customers’ affordability in that region, the price of the product, number of products sold online during last week and during last year during the same time as of the prediction time now, the level of online engagement with that product during last week and last year same time, number of product sold offline in stores during last week and last year during the same time as of the prediction time now, the level of online engagement with product during last week and last year same time, the price of the closest competitor product, the number of search queries related to that product during last week and the same time last year, and many other potential data features impacting the purchase of that product, and knowing how many of that product have been purchased within similar time, season, location or given context in past and classifying those into buckets/classes  for example Class A (demand lower than 5000 products), Class B (demand between 5001 and 10000 products), Class C (demand between 10001and 15000 products), Class D (demand between 15001 and 20000 products), and Class D (demand between 20001 and 25000 products) and Class E (demand above 25000 products),  predict the demand for that product (i.e. within which class/bucket the demand for that product will fall in (also calculate each band's prediction certainty (probability)).

If you wish to predict the demand as continuous numbers rather than categories or classes, you may predict the winning bucket as winning class of the prediction process and then divide that bucket into further M more granular sub-buckets and perform further rounds of training and prediction across the range of the winning bucket to get the winning sub-bucket and take the average value of that sub-bucket as the final continuous output value of the prediction. Please see the overview page for more details regarding the prediction with continuous numbers.       

2. Predict the next Black Friday sales (purchase amount of certain category of products for certain customers) and order and optimize Inventory/supply accordingly: Having the data features of thousands of past Black Friday purchases, including customer ID, product ID, gender of person who purchased it, age, history of black Friday purchases, occupation, city of purchase, number of years in current city, marital status, product category purchased, income, macroeconomics, inflation, customer trends and preferences, product brand popularity, and many other relevant data features and knowing the amount of purchase (as of for example five bands of purchase amount) as target prediction value, predict the purchase amount band (for example band 1, 2, 3, 4, 5) for that customer and for that category of product for next black Friday (also calculate each class's prediction certainty (probability)).

3. Predict the back order: A back order is an indicator of excessive customer demand for a certain product that leads to company’s capacity and supply shortage. This is usually result of strong sales performance (e.g., the product is in such high demand that production cannot keep up with sales). Back orders can lead to customer being upset, drop in their loyalty and even cancellation of some orders. Knowing the back order history of thousands of past product purchases including product sku, national inventory, lead time, quantity in transit, sales of past month, sales of past three months, sales of past six months, sales of past nine months, potential issues with product in past, and other data features relevant to back order and knowing that the product went on back order or not in past during a target season for example Christmas, predict whether another similar new product or the same product will go to back order during the next Christmas (and with what probability) and adjust the supply/inventory accordingly.

4. Package billing classification to help e-commerce companies reduce delivery costs: During past years the carriers have realised that cost models based on actual package weight don’t make sense for lightweight packages that take up cargo space and reduce the carrier’s shipment capacity, therefore they introduced billable weight. Knowing the package data features such as billable weight type, billable weight, weight, length, width, height, volume, dimensional weight, dimensional factor, and other relevant data features and knowing which price zone (The label and target of classification) those past packages ended up in, classify a new package bill into the correct pricing zone (and with what probability).

Or any other binary or multi-class supply chain-related classification or prediction use cases.


                                    Log in/Register                   Request a Demo                Instruction Manual