Lecture 1 - Patterns!
Patterns are fuel for a Data scientist!
Patterns Help us predict trends
Trend
A trend is the overall direction or consistent change observed in a repeating pattern over
time
Outliers
a data point that is very different from the others
ex:
Imagine your class took a math test, and most students scored between 60 and 90 marks,
but one student got 0 marks and another got 100 marks. These two are outliers because
they are very different from the rest.
Anomaly
Something that doesn't follow the expected pattern
ex: Bank Fraud Detection: If a person usually spends 1,000 per day but suddenly spends
1,00,000, the bank will suspect fraud.
Types of Patterns
1. Seasonal Patterns
2. Cyclical Patterns
3. Correlation Patterns - means two things changing together, one increase other also
increase (positive) or if one increase then one decrease (negative)
4. Clustering Patterns - means grouping things together,
i. ex: Netflix groups users based on movies they like
ii. ex: Kids with same interests tend to form groups with each other
5. Time series pattern
6. cause - effect pattern
7. frequency pattern
8. Fractal Patterns -
Power Law Distribution
Definition: A power law distribution follows the 80/20 rule, meaning a small number of
factors contribute to the majority of an effect.
Real-World Example : Internet Traffic (Pareto Principle): 80% of website traffic comes from
20% of users.
1. Wealth Distribution: A few people (1%) control most of the world's wealth.
2. Search Engine Rankings (Zipf's Law): The most searched words appear far more
frequently than less common words.
Technological Use Case:
1. Cybersecurity: Identifying the top 1% of network nodes responsible for most traffic to
detect DDoS attacks.
2. Social Media Influence: Platforms like Instagram and Twitter use power laws to rank
influencers.
Fractal Patterns in Data (Self-Similarity)
Definition: A fractal pattern is a self-repeating structure at different scales.
Real-World Example:
Stock Market Trends: Price movements often look similar over different time scales.
Neural Networks: Brain neurons form self-similar fractal structures.
Technological Use Case:
Al Image Compression: Fractal algorithms compress high-resolution images without losing
quality.
Telecommunications: Used in designing cellular network antennas for better signal strength.
Swarm Intelligence patterns
Definition: refers to a computational approach that leverages the collective behavior of
natural systems like ant colonies or bird flocks to solve complex problems by mimicking their
decentralized, self-organized decision-making process, often used for optimization tasks
where large datasets need to be analyzed and patterns identified without a central control
point.
Fourier Transform Patterns (FFT)
Definition: Converting complex time-based signals into frequency based patterns
Examples:
Heart rate monitoring (ECG)- detection of hear beats into waves on chart(Frequency tones)
Audio compression - MP3 files use FFT to reduce file size without losing qualities
Chaos Theory (Butterfly effect)
Definition: Tiny changes can create vastly different outcomes!!
Real world examples:
Weather Prediction: A small temp change can lead to storms
Tools:
PatternShapes -
use pattern shapes to generate the sequence
Julius AI basics - [Link]
Shows step wise solution to a problem
Kahoot - [Link]
online quiz thingy
Pattern Detective
game to find pattern in things no one can see! like finding patterns in the nooks where no
one will expect there to be a pattern there.
Homework:
PatternShapes Pro Tip: Create "shape + number" hybrid patterns (e.g., 2, 4, 6, )
Julius Al Advanced: Solve multi-step problems (e.g., "If rent increases by 10%, what's the
new total expense?").
Activity: "Predict the Future":
Use Julius Al to solve: "A pizza shop sells 30 pizzas daily. How many in July?"
"How could this help a business owner?"
New Words:
1. Transducer -
1. a device for producing an electrical signal from another form of energy such as pressure
2. a device that converts a one form of energy into another form of energy
2. Cryptography - the process of hiding information so that only the person a message was
intended for can read it