– Sharding vs. Replication: Sharding involves storing only portions of data on different servers, while replication involves duplicating the entire dataset on multiple servers.
– Hyperparameters in Ensemble Technique: Various hyperparameters are discussed, including the number of estimators (trees), learning rate, and gamma value for minimum split loss.
– Difference in Weightage in Gradient Boosting Algorithms: Extreme Gradient Boosting adjusts weightage given to different iterations based on decreasing information in residuals.
– Maximum Depth and Gamma Value: Maximum depth is a known parameter, while gamma represents the minimum split loss required for further splitting in tree-based models.
– Minimum Child Weight and Other Hyperparameters: Other hyperparameters mentioned include minimum sum of weights required for a node to split, subsamples, column samples, and various scaling factors.
#datascience #ensemble #technique #dataanalytics #bharanikumar
[ad_2]
source
