It predicts outcomes based on input data

It requires labeled data

It is used for reinforcement learning

It finds hidden patterns in data

It is always accurate

It requires a large amount of data

It does not require human input

It can achieve superhuman performance

A collection of data arranged in some order

A collection of algorithms

A method of data cleaning

A type of machine learning model

Data analysis

Model evaluation

Model deployment

Data gathering

To collect data from various sources

To analyze data for patterns

To clean and convert raw data into a usable format

To deploy models in production

No data

Random data

Labeled data

Unlabeled data

To classify data into categories

To discover patterns in data

To learn actions based on feedback

To clean and prepare data

K-means clustering

Logistic regression

Hierarchical clustering

Principal component analysis

The complexity of the model

The performance of a classification model

The amount of data collected

The accuracy of unsupervised learning

Association

Classification

Dimensionality Reduction

Clustering

A data structure

A continuous value

A numerical score

A categorical label

It is easier to implement

It can discover new patterns

It guarantees accurate results

It requires labeled data

Probabilistic Component Analysis

Pattern Classification Analysis

Predictive Classification Algorithm

Principal Component Analysis

Confusion matrix

Feature engineering

Data wrangling

Data gathering

Simple recall of facts

Applying skills and concepts

Reasoning and planning to solve problems

Collecting data

To predict outcomes

To group similar objects

To clean data

To classify data

It can yield different results with each run

It is always accurate

It is easy to implement

It requires labeled data

A model that is too simple

A model that learns too much from the training data

A model that performs well on unseen data

A model that is easy to interpret

Unsupervised learning requires more data

Supervised learning is easier to implement

Unsupervised learning is more accurate

Supervised learning uses labeled data, unsupervised does not

To predict outcomes based on input data

To clean data

To analyze data

To gather data

To classify data

To learn from feedback and rewards

To discover patterns in data

To clean and prepare data

Ignoring validation data

Using more features

Increasing model complexity

Data augmentation

Analyze Data

Develop

Get Data

Deploy

Multi-class Classifier

Regression Classifier

Binary Classifier

Cluster Classifier

K-Nearest Neighbours

Naïve Bayes

Logistic Regression

Decision Tree Classification

The hidden patterns in data.

The boundary that separates classes by as wide a margin as possible.

The closest neighbors in a training set.

The optimal number of clusters.

That the value of a particular feature is independent of the value of other features, given the output variable.

That the data must have linear decision boundaries.

That the algorithm should find the widest margin possible.

That the model must be trained on labeled data.

They consist of a single decision tree.

They subdivide feature space into regions with different labels.

They contain a large number of individual decision trees that operate as an ensemble.

They primarily use Bayes theorem of probability.

Classification Accuracy

Confusion Matrix

F1 Score

Precision and Recall

When the model is performing well.

When data is imbalanced.

When the output variable is a category.

When dealing with binary classifiers.

Predicting positive class as positive.

Predicting negative class as positive.

Predicting positive class as negative.

Predicting negative class as negative.

How good our model is at correctly predicting positive classes.

How many of the predictions are correct.

How good our model is when the prediction is positive.

The total variations within a cluster.

Random Operations Curve

Receiver Operating Characteristics Curve

Result Optimization Curve

Recall Over Classification Curve

To produce an accurate result for labeled data

To find the underlying pattern of the dataset and group data

To learn actions by trial and error

To predict specific outcomes

Classification

Regression

Clustering

Supervised Learning

Predicting labels for new data

Classifying data into predefined categories

Clustering n objects into k groups based on attributes

Evaluating model accuracy

To classify data into binary categories

To find hidden patterns in labeled datasets

To map original data into a lower dimensional space while preserving information

To identify fraudulent transactions

It always produces more accurate results

It is used for more complex tasks as it doesn't have labeled input data

It is intrinsically less difficult

It is easier to get labeled data for

Because it has corresponding output data

Because input data is labeled, and algorithms know the exact output

Because input data is not labeled, and algorithms do not know the exact output in advance

Because it is limited to simpler tasks

When the model performs poorly on training data

When the model cannot learn anything from the data

When the model learns noise and patterns specific to training data, reducing generalization

When the model performs better on unseen data than on training data

When the model performs extremely well on training data but poorly on test data

When the model cannot capture the patterns in training data

When the model is too complex

When the model uses too much data

Increase the number of training epochs

Remove dropout layers

Apply data augmentation

Decrease model complexity

To increase the learning rate

To prevent underfitting

To randomly remove data points

To reduce overfitting by preventing co-adaptation of neurons

Apply more regularization

Add dropout layers

Decrease the number of training iterations

Increase the complexity of the model

Decreasing regularization

Removing dropout layers

Using a simpler model

Increasing training epochs

Requires manual rule-setting by experts

Cannot outperform humans in any task

Learns patterns automatically from data

Always produces simple and explainable models

They use too little data

They require human-written rules

They can be easily tricked by poor-quality data

They are always better than humans in all tasks

Models can explain all of their decisions clearly

Machine learning requires very little data to be accurate

The relationships learned are often complex and not easily interpretable

Machine learning is unsuitable for image recognition tasks

It requires no data to train

It can learn and optimize strategies beyond human capability

It only works on text-based tasks

It avoids using past data completely