Simple prediction rule
A very simple 'model' is a rule we write by hand. For example: if score >= 80 then predict 'pass'. Machine learning later learns such rules from data. Here we write a small rule and use it on a few examples.
Appy Says…
A spam filter doesn't know what spam is — it's learned from millions of examples. A price predictor doesn't have a formula — it's found patterns in data. Simple prediction is AI's simplest superpower: look at past data, predict future values.
What is Simple Prediction in AI?
Prediction is using patterns in past data to output a value for new, unseen inputs. The simplest form is linear regression — finding the line that best fits a set of data points.
- •Input features → model → predicted output
- •Regression: predicts a continuous number (price, temperature, score)
- •Classification: predicts a category (spam/not-spam, cat/dog)
- •Training: showing the model labelled examples to find the pattern
- •Prediction: giving the trained model a new input to get an output
- •Accuracy: tested on data the model hasn't seen (test set)
Think of it like predicting a Minecraft biome from coordinates
If you plot a thousand Minecraft coordinates against their biome and temperature, you'd spot patterns — certain coordinates tend to be cold, others warm. A prediction model finds these patterns mathematically and generalises them to predict any new coordinate's climate.
How It Works
- •1. Collect labelled data: (input, correct output) pairs
- •2. Split: 80% training data, 20% test data
- •3. Train: model adjusts its parameters to minimise prediction error on training data
- •4. Evaluate: measure error on test data (data the model hasn't seen)
- •5. Predict: pass new input through the trained model → get prediction
- •6. Common measure of error: Mean Squared Error (MSE) or accuracy %
Real-World Examples
- •Spotify: predicts how much you'll like a song based on listening history
- •Netflix: predicts your star rating for unwatched films
- •Amazon: predicts estimated delivery time based on location + carrier history
- •School: predicting exam performance from homework completion rate
Key Facts
- •Linear regression is the simplest ML algorithm — it fits a straight line to data
- •Decision trees and random forests are easy-to-understand models for classification
- •Overfitting: model memorises training data but fails on new data — too complex for the dataset
- •More data almost always beats a more complex model for improving prediction accuracy
Watch Out!
Never evaluate a model on the same data you trained it with — it will appear to have perfect accuracy (it just memorised the answers). Always hold back a test set. If test accuracy is much lower than training accuracy, the model is overfitting.
Remember
Prediction = pattern from past data applied to new inputs. Train on labelled data, test on unseen data. Regression predicts numbers; classification predicts categories.
What You Learned
- •AI prediction: find patterns in labelled training data, apply to new inputs
- •Train/test split prevents overfitting; evaluate on data the model never saw
- •Unlocks: understanding how every recommendation system and classification AI works
Key Facts
- →Linear regression is the simplest ML algorithm — it fits a straight line to data
- →Decision trees and random forests are easy-to-understand models for classification
- →Overfitting: model memorises training data but fails on new data — too complex for the dataset
- →More data almost always beats a more complex model for improving prediction accuracy
Real-World Examples
Remember
Prediction = pattern from past data applied to new inputs. Train on labelled data, test on unseen data. Regression predicts numbers; classification predicts categories.
Quick Quiz
What is a simple prediction rule?