{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/f3d2e099873d4a0f8c0ca91e1d369900\" frameborder=\"0\" width=\"1920\" height=\"1440\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1440,"width":1920,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1440,"thumbnail_width":1920,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/f3d2e099873d4a0f8c0ca91e1d369900-4e387084f321250e.gif","duration":166.494,"title":"Feature Engineering","description":"In this video, I discuss the importance of using weight of evidence for categorical variables in fraud detection. I explain how this method helps us avoid overfitting by providing a numerical encoding associated with fraud ratios. It's crucial that we perform these calculations solely on the training set to prevent data leakage. I also emphasize the need to maintain the integrity of unseen data for accurate model performance. Please take note of this approach as we move forward."}