Revision Session 1

Revision Session 1

1. AI & ML

   • What is the difference between AI & Machine Learning ?
     • AI is a broad field that involves creating systems capable of performing tasks that require human intelligence, such as reasoning, learning, and problem-solving. Machine Learning is a subset of AI that focuses on enabling systems to learn from data and improve their performance.
   • Popular AI Progammes
     • Google Gemini
     • Grammarly
     • Tesla AutoPilot
   • How can AI go wrong ?
     • AI can go wrong in several ways, such as making biased decisions if trained on skewed data, leading to unfair outcomes. It can also misinterpret human instructions, resulting in unintended actions. Security vulnerabilities in AI systems can be exploited, causing harm or data breaches.
   • Example of the last time I used Ai
     • Asking Google Assistance for horror film recomendations on Netflix
   • What is an Algorithm
     • An algorithm is a set of step-by-step instructions designed to solve a specific problem or perform a task.
   • How can Machine Learning be biased?
     • Machine learning can be biased due to issues in data or algorithms. Selection bias happens when training data isn’t representative, leading to poor performance on underrepresented groups. Algorithmic bias comes from the design of the algorithm itself. Confirmation bias occurs when data reinforces stereotypes, like a hiring algorithm favoring men if trained on biased data. Exclusion bias arises when important variables or groups are missing.

2. Orange Data Mining

   • Label
     • A label is an output ot target variable that the model is trained to predict.
   • Machine Learning Model
     • Clustering
     • Classification
     • Regression
       • A regression model is a type of supervised learning algorithm used to predict continuous values by estimating the relationship between a dependent variable and one or more independent variables.
   • Feature
     • A feature is an individual measurable property or characteristic of the data used as an input variable for the algorithm.
   • Supervised Model
     • A supervised model in machine learning is trained on labeled data, where each input is paired with the correct output. It learns by comparing its predictions to the actual labels and adjusting to minimise errors.
     • Example - A dataset trained on emails being able to categorise as "Spam" or "not Spam".
   • Unsupervised Model
     • An unsupervised model in machine learning is an algorithm that is trained on data without labeled responses.
     • K -means Clustering groups data based on similarities.

3. Orange Data Mining Terminology

   • Testing Split
     • A testing split involves dividing a dataset into training and testing subsets.
       • Evaluate Performance = Seek Improvements for accuracy.
   • Training Split
     • This learns patterns and relationship with data allowing to evaluate the performance of unseen data.
   • Scoring
     • Involves calculating metrics that quantify how well a model performs.
   • Evaluating
     • This is cross-validation where the data splits into training and testing multiple times to ensure its generalised and reliable.

4. What does scoring include?

   • Accuracy
     • Correct predictions divided by the total number of predictions i.e. (TP + TN)/Total predictions made
   • Precision
     • The fraction of the cases classified as positive that are actually positive i.e. TP/(TP + FP)
   • Recall
     • The fraction of positive cases correctly identified i.e. TP/(TP + FN).
   • F1
     • A metric that combines both precision and recall

5. Classification Model

   • What is a classification model
     • A classification model predicts categorical labels for a given input. For example, in binary classification, it classifies data points into two categories such as 1 (positive) or 0 (negative). The outcome from a classification model includes predicted labels and associated probabilities.
   • Example of how an classification model is used
     • In medical diagnosis is to determine whether a patient has a disease (positive) or not (negative).

6. Confusion Matrix

   • Relative Squared Error
     • A relative metric between 0 and 1 based on the square of the differences between predicted and true values.
   • Coefficient of Determination
     • A relative metric between 0 and 1 based on the absolute differences between predicted and true values.
   • Root Mean Squared Error
     • The square root of the mean squared difference between predicted and true values
   • Mean Absolute Error
     • The average difference between predicted values and true values.
   • Relative Absolute Error
     • Determines what proportion of error is explained by the model.