𝗖𝗮𝘂𝘀𝗮𝗹 𝗔𝗜 𝗘𝘅𝗽𝗹𝗮𝗶𝗻𝗲𝗱: 𝗗𝗮𝘆 𝟱 - 𝗖𝗼𝘂𝗻𝘁𝗲𝗿𝗳𝗮𝗰𝘁𝘂𝗮𝗹𝘀 We’ve covered the do-operator, the trio of confounders, colliders, and mediators, and interventions. Now, let’s explore the "what could have been" with 𝗰𝗼𝘂𝗻𝘁𝗲𝗿𝗳𝗮𝗰𝘁𝘂𝗮𝗹𝘀: the ultimate "what if" of Causality. Interventions ask, "What happens if we change X?". Counterfactuals dig deeper: "Given what happened, what would’ve occurred if X had been different in the past?". They imagine alternate realities from real data, revealing insights beyond standard ML. 𝗛𝗼𝘄 𝗜𝘁 𝗪𝗼𝗿𝗸𝘀 Picture a customer who didn’t buy insurance after a 10% discount offer. A counterfactual asks: "Would they have bought it if we’d offered 20% instead?" Mathematically, it’s about reconstructing observed data. If Y is the outcome (e.g., purchase) and X is the treatment (e.g., discount), we estimate Y under do(X = 20%) for a case where X = 10% occurred, adjusting for confounders like income or age. Using a causal model, we estimate their behaviour in this alternate scenario, accounting for their profile and historical trends. 𝗥𝗲𝗮𝗹-𝗪𝗼𝗿𝗹𝗱 𝗘𝘅𝗮𝗺𝗽𝗹𝗲 Consider a clinical study: Patients who received a vaccine didn’t show reduced symptoms. The counterfactual asks: "Would symptoms have decreased if they’d gotten a higher dose?". A model might estimate a 30% symptom reduction in that alternate scenario, revealing the dose’s potential impact. This isn’t just predicting, it’s rewriting the past. 𝗪𝗵𝘆 𝗜𝘁’𝘀 𝗣𝗼𝘄𝗲𝗿𝗳𝘂𝗹 • Hindsight: Understands what could’ve worked, not just what did. • Personalisation: Tailors insights to individual scenarios (e.g., "What if this patient got treatment X?"). • Strategy: Informs future decisions with lessons from imagined pasts. 𝗧𝗮𝗸𝗲𝗮𝘄𝗮𝘆: 𝗖𝗼𝘂𝗻𝘁𝗲𝗿𝗳𝗮𝗰𝘁𝘂𝗮𝗹𝘀 𝗮𝗿𝗲 𝗖𝗮𝘂𝘀𝗮𝗹 𝗔𝗜’𝘀 𝘁𝗶𝗺𝗲 𝗺𝗮𝗰𝗵𝗶𝗻𝗲, 𝗿𝗲𝘃𝗲𝗮𝗹𝗶𝗻𝗴 𝗮𝗹𝘁𝗲𝗿𝗻𝗮𝘁𝗲 𝗼𝘂𝘁𝗰𝗼𝗺𝗲𝘀 𝘁𝗼 𝘀𝗵𝗮𝗿𝗽𝗲𝗻 𝗿𝗲𝗮𝗹-𝘄𝗼𝗿𝗹𝗱 𝗰𝗵𝗼𝗶𝗰𝗲𝘀. Have you ever thought about alternative scenarios to help making a business decision? Share below! 𝘐𝘮𝘢𝘨𝘦 𝘤𝘳𝘦𝘥𝘪𝘵: 𝘛𝘩𝘦 𝘍𝘢𝘪𝘳𝘯𝘦𝘴𝘴 𝘍𝘪𝘦𝘭𝘥 𝘎𝘶𝘪𝘥𝘦: 𝘗𝘦𝘳𝘴𝘱𝘦𝘤𝘵𝘪𝘷𝘦𝘴 𝘧𝘳𝘰𝘮 𝘚𝘰𝘤𝘪𝘢𝘭 𝘢𝘯𝘥 𝘍𝘰𝘳𝘮𝘢𝘭 𝘚𝘤𝘪𝘦𝘯𝘤𝘦𝘴, 𝘊𝘢𝘳𝘦𝘺 𝘦𝘵 𝘢𝘭. 2022 ( 𝘩𝘵𝘵𝘱𝘴://𝘸𝘸𝘸.𝘳𝘦𝘴𝘦𝘢𝘳𝘤𝘩𝘨𝘢𝘵𝘦.𝘯𝘦𝘵/𝘱𝘶𝘣𝘭𝘪𝘤𝘢𝘵𝘪𝘰𝘯/357875366_𝘛𝘩𝘦_𝘍𝘢𝘪𝘳𝘯𝘦𝘴𝘴_𝘍𝘪𝘦𝘭𝘥_𝘎𝘶𝘪𝘥𝘦_𝘗𝘦𝘳𝘴𝘱𝘦𝘤𝘵𝘪𝘷𝘦𝘴_𝘧𝘳𝘰𝘮_𝘚𝘰𝘤𝘪𝘢𝘭_𝘢𝘯𝘥_𝘍𝘰𝘳𝘮𝘢𝘭_𝘚𝘤𝘪𝘦𝘯𝘤𝘦𝘴 ) #CausalAI #DataScience #Causality #ArtificialIntelligence #AIExplained #Counterfactuals
