In the explanation about how the decision tree is made, the lecturer said that starting from 98988 samples, the first split send 76705 and 22283 samples to the left and right, respectively. Then I suppose the next split on the left with use 76705 sample that have been sent. However, on the tree plotted, this left split works on 82418 samples which makes me confused. The similar patterns have been applied for other splits. Would any one help me to explain it? Thanks a lot!