NEW
Font size
WorksheetsReview of Machine Learning
Total questions: 23
Worksheet time: 12mins
What is Machine Learning?
The process of teaching computers to perform tasks without explicitly programmed instructions.
The study of algorithms that can improve automatically through experience.
Machine learning allows computers to learn and infer from data.
The process of teaching computers to perform tasks using physical movements and sensors.
Types of Machine Learning:
Supervised Learning and Unsupervised Learning
Deep Learning, Natural Language Processing, and Computer Vision
Genetic Algorithms, Neural Networks, and Decision Trees
Reinforcement Learning, Dimensionality Reduction, and Clustering
In Supervised Data points have:
known outcome
a category
continuous (numerical)
unknown outcome
In unSupervised Data points have:
known outcome
unknown outcome
continuous (numerical)
a category
In regression Outcome is:
have known outcome
continuous (numerical)
unknown outcome
a category
In classification Outcome is
continuous (numerical)
unknown
known
a category
Target:
the target value for a single data point
properties of the data used for prediction (non-target columns)
predicted category or value of the data (column to predict)
a single data point within the data (one row)
Features:
predicted category or value of the data (column to predict)
properties of the data used for prediction (non-target columns)
a single data point within the data (one row)
the target value for a single data point
Example:
predicted category or value of the data (column to predict)
properties of the data used for prediction (non-target columns)
a single data point within the data (one row)
the target value for a single data point
Label:
predicted category or value of the data (column to predict)
properties of the data used for prediction (non-target columns)
a single data point within the data (one row)
the target value for a single data point
Target:
Response, Output, Dependent Variable, Labels
Predictors, Input, Independent Variables, Attributes
Observation, Record, Instance, Datapoint, Row
Answer, y-value, Category
Features:
Response, Output, Dependent Variable, Labels
Predictors, Input, Independent Variables, Attributes
Observation, Record, Instance, Datapoint, Row
Answer, y-value, Category
Example:
Response, Output, Dependent Variable, Labels
Predictors, Input, Independent Variables, Attributes
Observation, Record, Instance, Datapoint, Row
Answer, y-value, Category
Label:
Response, Output, Dependent Variable, Labels
Predictors, Input, Independent Variables, Attributes
Observation, Record, Instance, Datapoint, Row
Answer, y-value, Category
Types of Classification Predictions: Hard Prediction:
Predict a single category for each instance.
Rank the instances from most likely to least likely. (binary classification)
Assign a probability distribution across the classes to each instance.
Types of Classification Predictions: Ranking Prediction:
Predict a single category for each instance.
Rank the instances from most likely to least likely. (binary classification)
Assign a probability distribution across the classes to each instance.
Types of Classification Predictions: Probability Prediction:
Predict a single category for each instance.
Rank the instances from most likely to least likely. (binary classification)
Assign a probability distribution across the classes to each instance.
Metrics for Classification: Hard Prediction:
Accuracy, Precision, Recall (Sensitivity), Specificity, F1 Score
AUC (ROC), Precision-Recall Curves
Log-loss (aka Cross-Entropy), Brier Score
Metrics for Classification: Ranking Prediction:
Accuracy, Precision, Recall (Sensitivity), Specificity, F1 Score
AUC (ROC), Precision-Recall Curves
Log-loss (aka Cross-Entropy), Brier Score
Metrics for Classification: Probability Prediction:
Accuracy, Precision, Recall (Sensitivity), Specificity, F1 Score
AUC (ROC), Precision-Recall Curves
Log-loss (aka Cross-Entropy), Brier Score
Metrics for Regression:
Accuracy and Precision
Mean Squared Error (MSE) and F1 score
Root Mean Square Error (RMSE) and Mean Absolute Deviation (MAD)
R-squared and Recall
Using Training and Test Data: Training data:
The training data is used to validate the model's performance.
The training data is used to assess the model's generalization to unseen data.
to fit the model
The training data is used to fine-tune the hyperparameters of the model.
Using Training and Test Data: Test data:
The test data is used to build and train the model.
The test data is used to validate the model's performance.
measure performance (predict label with model/ compare with actual value /measure error)
The test data is used to fine-tune the hyperparameters of the model.
