48) Amazon QuicksightAn ML-powered business intelligence tool that can be integrated with Amazon Athena for creating visualizations Nov 7, 2023, 5:07 PM Navid
49) Amazon Quicksight ML InsightsCan produce forecasts on time-series data without requiring prior machine learning knowledge Nov 7, 2023, 5:07 PM Navid
54) One-Hot EncodingBest suitable for nominal data where categories are transformed into a binary matrix Nov 7, 2023, 5:07 PM Navid
55) T-SNE GroupingReduces dimensions by grouping similar data points Nov 7, 2023, 5:07 PM Navid
65) High VarianceIndicates overfitting where the model is too complex and does not generalize well to new data Nov 7, 2023, 5:07 PM Navid
66) Batch NormalizationOvercomes vanishing and exploding gradient problems in neural networks Nov 7, 2023, 5:07 PM Navid
74) Sigmoid Activation FunctionUsed for binary classification to predict the probability of each class independently Nov 7, 2023, 5:07 PM Navid
46) Right-Skewed DataWhen the graph has a long right tail it is right-skewed and log transform can help in making the data more normally distributed Nov 7, 2023, 5:07 PM Navid
47) SMOTESynthetic Minority Oversampling Technique used to increase the number of minority class instances in a dataset Nov 7, 2023, 5:07 PM Navid
58) Handling OutliersApproaches include removal transformation such as log transform or keeping them if they hold valuable information Nov 7, 2023, 5:07 PM Navid
56) Data Labeling OptionsInclude Mechanical Turk private labeling workforce and vendors for third-party services Nov 7, 2023, 5:07 PM Navid
57) ScalersStandardize the range of data features Nov 7, 2023, 5:07 PM Navid
67) RelU Activation FunctionDoes not have the vanishing gradient problem and adds non-linearity to the output of a neuron Nov 7, 2023, 5:07 PM Navid
69) BaggingGenerates new training sets by random sampling with replacement and can train models in parallel Nov 7, 2023, 5:07 PM Navid
68) OptimizersInclude Adam Nov 7, 2023, 5:07 PM Navid
78) AUC of ROC CurveEvaluates how well a model distinguishes between classes with a higher AUC indicating better performance Nov 7, 2023, 5:07 PM Navid
76) L2 RegularizationA regularization method that keeps all features but penalizes weights to prevent overfitting Nov 7, 2023, 5:07 PM Navid
43) Decision TreesFeature selection is embedded in decision trees which makes them less impacted by strongly correlated features Nov 7, 2023, 5:07 PM Navid