Cross-Validation in ML:

What is Cross-Validation?

[{"selector":"#anim-6290a9c5-53f0-4c3d-addc-fd2745aeffd6","keyframes":{"opacity":[0,1]},"delay":500,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-74b09c8d-20d4-49b9-92dd-c81d667e10d0","keyframes":{"opacity":[0,1]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-93f41f78-e14a-4d06-beee-fbf390047d27","keyframes":{"transform":["translate3d(-114.96816%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-89b5a85f-5b5e-4021-bea4-6c7b3ba1ba69","keyframes":{"opacity":[0,1]},"delay":500,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-9ef26d0a-5056-4f76-8737-02327f53bcad","keyframes":{"transform":["translate3d(0px, -143.76124%, 0)","translate3d(0px, 0px, 0)"]},"delay":500,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-6f61d5c5-7d0c-458f-9e78-9693a089d00f","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-872a7aed-1c8b-4ca1-88b7-083df715b4cb","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-417732b3-b416-4fd0-b272-8e9b8a1d30d2","keyframes":{"opacity":[0,1]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-2e2fd3d7-661a-449a-9f4d-a12c1bd0cc9b","keyframes":{"transform":["translate3d(0px, -287.08698%, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] Cross-validation estimates how well a ML model generalises to new, unseen data. Essential! Read More

Why Use It?

[{"selector":"#anim-947f84e7-bfa1-4caa-8f9c-911afd6789e1","keyframes":{"opacity":[0,1]},"delay":500,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-137708e8-b58a-4fa4-911a-5208897bb0b3","keyframes":{"opacity":[0,1]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-b1945e3b-cf9a-44ee-9a1a-2c60d489a094","keyframes":{"transform":["translate3d(-115.28663%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-0b775fb0-f133-462b-a6c3-ba6ea38bebdb","keyframes":{"opacity":[0,1]},"delay":500,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-ba744981-90bd-4d89-8506-6f7a2722870d","keyframes":{"transform":["translate3d(0px, -176.13928%, 0)","translate3d(0px, 0px, 0)"]},"delay":500,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-318e17d6-1843-4cea-8e96-35cd86aa54b3","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-09242636-40f4-457a-90b4-433affd329d6","keyframes":{"opacity":[0,1]},"delay":0,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-21b15a8f-9c06-4b78-b7b4-889737e66790","keyframes":{"opacity":[0,1]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-83c87055-941c-45cf-b3b5-9b5b010a8313","keyframes":{"transform":["translate3d(0px, -287.08698%, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] Avoid overfitting! CV gives realistic performance. Tune hyperparameters accurately. Read More

K-Fold CV Explained

[{"selector":"#anim-dbd7395f-9e8e-4670-a6e9-d249c85f3650","keyframes":{"opacity":[0,1]},"delay":500,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-4143ca1d-2e70-4993-984f-3108fa5c8fb6","keyframes":{"opacity":[0,1]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-680ff478-2604-4916-8062-a3afa7c21610","keyframes":{"transform":["translate3d(-115.18988%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-845c65da-73ec-4899-b0fc-6d5e8ef0a088","keyframes":{"opacity":[0,1]},"delay":500,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-bc6f3348-5d15-4c44-bc76-b6ec16ad6bad","keyframes":{"transform":["translate3d(0px, -181.66924%, 0)","translate3d(0px, 0px, 0)"]},"delay":500,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-2f20542b-4213-449a-b4e6-fb969467b2df","keyframes":{"opacity":[0,1]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-5eb5dcbf-d703-49bc-a38f-b0250b932bb6","keyframes":{"transform":["translate3d(0px, -287.08698%, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] Data split into K "folds." K-1 train, 1 test. Rotate! Average the results.

Common Type: 5-Fold CV

[{"selector":"#anim-419a778c-232f-463e-8a59-1fb651ed4eb3","keyframes":{"opacity":[0,1]},"delay":500,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-e2f2e4f2-21b9-46d1-be7a-6399c9543cdc","keyframes":{"opacity":[0,1]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-c4d03b0b-5673-40a1-a00a-2d70e602d404","keyframes":{"transform":["translate3d(-115.28663%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-9f9817f2-e3f3-4b41-9730-a4f3a95207d1","keyframes":{"opacity":[0,1]},"delay":0,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-0d2d7144-9433-4578-b2ca-fd74112cc97b","keyframes":{"transform":["translate3d(0px, -201.80014%, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-7444fafb-b721-4b8b-baaf-f9557d443430","keyframes":{"opacity":[0,1]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-0bffcb4a-73c7-45e4-a3ae-69bc08b4c1cf","keyframes":{"transform":["translate3d(0px, -287.08698%, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] Popular choice. Good balance: computation vs. reliability. Repeat several times.

Stratified K-Fold

[{"selector":"#anim-b7e30d37-fbe6-4cd3-9e21-04bdd3e56d38","keyframes":{"opacity":[0,1]},"delay":500,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-f4d5f1c6-97cb-4836-a7c7-bb49f0655a94","keyframes":{"opacity":[0,1]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-b431b5ac-ea40-4140-9350-5fa6220035a6","keyframes":{"transform":["translate3d(-115.92356%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-eb89ee8b-01b3-4f74-999b-757320db572f","keyframes":{"opacity":[0,1]},"delay":500,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-22dca57e-6a73-4ddf-8709-26cbcc3ea1f1","keyframes":{"transform":["translate3d(0px, -245.78675%, 0)","translate3d(0px, 0px, 0)"]},"delay":500,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-69acd7e2-549b-4bce-b27a-11954c513adb","keyframes":{"opacity":[0,1]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-a2e4ae03-b305-4362-ae5f-51f1873e6dd0","keyframes":{"transform":["translate3d(0px, -287.08698%, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] Ensures each fold has similar class distribution. Ideal for imbalanced datasets.

CV Avoids Data Leakage

[{"selector":"#anim-29b54757-d852-4cf3-a46b-ebd850da2feb","keyframes":{"opacity":[0,1]},"delay":500,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-9b0034ad-311c-4918-8176-09ed867ab6d2","keyframes":{"opacity":[0,1]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-64f4f56f-0892-4dfb-8a9f-458ed78c7ad9","keyframes":{"transform":["translate3d(-114.96816%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-bba8b8c5-af76-41fd-89f3-047b1cb0b2dc","keyframes":{"opacity":[0,1]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-b1618c9d-3115-4ec9-92f8-1575c9cd35e8","keyframes":{"transform":["translate3d(0px, -287.08698%, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-4e8fa94c-2925-41db-8a61-e311a04247c8","keyframes":{"opacity":[0,1]},"delay":500,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-49fdfdb9-41e5-4415-a0cb-b318e26c8906","keyframes":{"transform":["translate3d(0px, -177.57757%, 0)","translate3d(0px, 0px, 0)"]},"delay":500,"duration":1500,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] Key! Properly split data before feature scaling. Prevent biased evaluation!

Use Cross-Validation Now!

[{"selector":"#anim-4e9e5cd4-c3cb-4488-9de6-dbb75a097798","keyframes":{"opacity":[0,1]},"delay":500,"duration":600,"easing":"cubic-bezier(0.4, 0.4, 0.0, 1)","fill":"both"}] [{"selector":"#anim-9a6c87f9-380d-4f18-8ced-015ba349b4b4","keyframes":{"opacity":[0,1]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] [{"selector":"#anim-48333682-cf20-4430-a398-a0eb0b24a885","keyframes":{"transform":["translate3d(-120.70064%, 0px, 0)","translate3d(0px, 0px, 0)"]},"delay":0,"duration":700,"easing":"cubic-bezier(0.2, 0.6, 0.0, 1)","fill":"both"}] Essential for building robust, generalisable machine learning models. Start today!