How Explainable AI Can Elevate Your Decision Making Process
Transform your business with AI that doesn't just tell you "what" but also explains "why" - the factor to a successful business.
The Story That Started It All
Let me tell you a story. Once an economist saw that in his town, the ice cream sales were increasing but so were the number of people falling ill. He quickly called the food safety department and told them to test the only ice cream shop in town. They found no issues, the ice cream shop infact had the highest standards and quality of ingredients. The economist was left confused, and people called him an idiot.
He remembered a wise man once said "Correlation is not Causation"
He then set to understand the real cause behind the increasing number of ill people
Soon found that due to heat the food people stored at home rotted fast which caused the illness and not the ice cream
People enjoyed ice cream because it kept them cool in the summers
This is exactly why we need Explainable AI - to understand the "why" behind predictions
From Traditional ML to Explainable AI: A Real Business Example
Imagine you are a founder who runs a clothes brand and you are doing well. Getting a few hundred sales in your store and on through delivery platforms throughout the day. You have a good set of repeat customers as well who buy your products multiple times. So you now have a huge dataset of your customers and their information.
See what kind of customers are buying what product (Eg: Men from South bombay mostly buy black oversized T-Shirts of size M and XL, while women from Noida mostly buy Blue jeans of size 34 to 36)
See when your customers are buying the most (Eg: Just before a huge music event your sales blast as most of your customers are Gen-Z, while they are relatively less orders in months of July, probably due to high rains and less events)
Make strategic decisions: Increase paid ads for black oversized tees in other parts of mumbai, or have more staff in January and February due to Lollapalooza which happens in February
Go one step ahead and use ML Algorithm (Eg: XgBoost) to predict with 90% accuracy what customer is likely to buy your product again or not
But here's where Explainable AI changes everything...
How Explainable AI Transforms Your Insights
Traditional ML Prediction
The model predicts Customer 8234: will buy your product again. But what next? What will you do with this information?
Explainable AI Enhancement
The model predicts Customer 8234: will buy your product again with a confidence score of 80/100. As he is from Bandra and has visited your website multiple amount of times in past few days and also paid with Axis bank credit card. These metrics show a strong indication this customer to purchase from your store again.
Feature Importance Analysis
The model says that it considers "time spent on the website" the largest contributor for the prediction, followed by delivery time and location, age. This shows ExplainableAI goes beyond traditional Machine Learning and does not only tell you what, but also "why?"
Actionable Optimization
Simulate what is the optimum time a person spends on your website to buy the product. Once you find out the optimum time (say 7.5 minutes) you can sit with your UX team to redesign the website/make changes in such a way that each customer spend about that much time.
Counterfactual Analysis
Create "Counterfactuals" Eg: If customer 2847 from Pune spent 6 minutes instead of the 10 minutes he actually spent would he be more likely to buy? or if we promised 3 day delivery instead of 7, what effect would that have on the prediction?
Ready to Elevate Your Decision Making?
With the addition of ExplainableAI, the decision making process elevates and you can make good, more confident decisions to improve and scale your business. If you want to implement Analytics and ExplainableAI in your decision making process, reach out to us - we will help you create a strategy which you can use to win and get ahead of your competitors.